(navigation image)
Home American Libraries | Canadian Libraries | Universal Library | Community Texts | Project Gutenberg | Children's Library | Biodiversity Heritage Library | Additional Collections
Search: Advanced Search
Anonymous User (login or join us)
Upload
See other formats

Full text of "Major Interchange Design, Operation, and Traffic Control"

9rt No. FHWA-RD-73-81 



IBSpARTMiiN'i' Or 

trp;n°.^ortation 
JUN 1 1974 

LJLDXU-i-tiY 



lOR INTERCHANGE DESIGN, OPERATIOI 
) TRAFFIC CONTROL 



Vol. 2. Appendixes A - G 



J. I. Taylor and others 




^r^ni o* '■ 



July 1973 
Final Report 



This document is available to tiie public 
through the National Technical Information 
Service, Springfield, Virginia 22151 



Prepared for 

FEDERAL HIGHWAY ADMINISTRATION 

Offices of Research & Development 
Washington. D.C. 20590 



NOTICE 

This document is disseminated under the sponsorship 
of the Department of Transportation in the interest 
of information exchange. The United States 
Government assumes no liability for its contents or 
use thereof. 

The contents of this report reflect the views of the 
contracting organization, which is responsible for 
the facts and the accuracy of the data presented 
herein. The contents do not necessarily reflect the 
official views or policy of the Department of 
Transportation. This report does not constitute a 
standard, specification, or regulation. 



7S' 



TECHNICAL REPORT STANDARD TITLE PAGE 



1. Keport No. 

FHWA-RD-73-81 



2. Government Accession No. 



3, Recipient's Catalog No, 



4. Title ond Subtitle 

^JOR INTERCHANGE DESIGN, OPERATION, AND TRAFFIC 
CONTROL 
Vo 1 • 2 . Appendixes A-G 



5. Report Date 

■Tiilv 197 3 



6. Performing Organization Code 



7. Author(s) 

J.I. Taylor, R.A. Olsen, J.C. Hajrward, W.L. Raymond, Jr 
and R.S. Hostetter 



8. Performing Organi zotion Report No. 

TTSC 7315 



9. Performing Organization Nome ond Address 

Pennsylvania Transportation and Traffic Safety Center 

Research Bldg. B 

University Park, Pennsylvania 16802 



10. Work Unit No. 

FCP 31P2-502 



11. Contract or Grant No. 

DOT-FH-ll-7795 



12. Sponsoring Agency Name and Address 

Offices of Research and Development 
Federal Highway Administration, 
U.S. Department of Transportation 
Washington, D.C. 20590 



13. Type of Report and Period Covered 

Final Report 



14. Sponsoring Aqency Code 

T-0080 



15. Supplementory Notes 

FHWA contract manager: C. R. Stockfisch (HRS-33) 

For table of contents of this report, see Vol. 1 (FHWA-RD-73-80) . 



16. Abstract ■j^g objectives of this research project were to develop improved design pro- 
cedures and guidelines for major (i.e., freeway-to-freeway) interchanges through the 
examination and analysis of existing design procedures and current freeway operational 
characteristics. Pertinent information was gathered through a review of the 
literature, conversations with and workshop participation by practicing design 
engineers and traffic operations specialists, and through written questionnaires. 

The criteria and guidelines used in the design of major interchanges at both 
the overall configuration level and the individual component levels (such as 
entrances, exits, lane drops, major forks) are reviewed; conclusions and recommenda- 
tions for future practices are stated. Freeway traffic control systems are examined 
in the context of major interchange design and operation, and- the implications of 
various systems are explained. 

A methodology for interchange evaluation using decision theory and tradeoff 
analyses is presented, with example applications. Extensive case studies of a lane 
drop and exit ramps at a major interchange are described to illustrate the manner in 
which the recommended guidelines might be applied in practice. Two sample "Fact 
Sheets" illustrating the manner in which design experience information might be dis- 
seminated to the design community are included. 

A bibliography of over 200 pertinent references accompanies the report. This 
report is in three volumes. The other volumes are: 

FHWA-RD-73-80. Vol. 1. Text of Report 
FHWA-RD-73-82. Vol. 3. Appendixes H-M 



17. Keywords Ma j or Interchange, Freeways, 
Freeway Control, Highway Design, Freeway 
Operations, Design Policy, Design 
Procedures, Decision Theory 



18, Distribution Statement 



No restrictions. This ciocument is 
available to the public through the 
National Technical Information Service, 
Springfield, Virginia 22151. 



19. Security C.lassif. (of this report) 

Unclassified 



20. Security Classif. (of this page) 

Unclassified 



21. No. of Pages 
1251 



22. Price 



Form DOT F 1700.7 (e-ss) 



"bEPARtlVE:^!' OF 
TRANSPORTATION 

JUN 1 1974 

UBRARY 



APPENDIX A 
SUMMARIES OF INTERIM REPORTS 






A-i 



APPENDIX A: SUMMARIES OF INTERIM REPORTS 

Introduction 

In the course of the project, three interim reports were published 
under separate cover. The summaries of these three reports are included 
in this appendix to indicate the nature of the information available 
therein. 

Interim Reports 2 and 3 are essentially independent, whereas much 
of the material contained in Interim Report 1 has been incorporated 
in this report. 

Interim Report //I; Current Practices and Research Review 
General Findings 

The topic of major interchanges is rarely treated as a separate 
entity either by state highway officials or by researchers. More often 
than not, major interchanges are coupled with minor interchanges, both 
of which are considered under the general topic of grade-separated inter- 
sections. Indeed, it may be that it is this failure to recognize the 
special problems and unique circumstances associated with freeway-to- 
freeway connections that results in the less-than-ideal traffic flows 
which characterize some major interchanges. 

In the matter of design procedures, each state has its own, and 
seems reluctant to consider adopting a nationwide or regional standard. 
State highway officials argue that their procedures match their own 
requirements and regional peculiarities and that a standard procedure 
for all states would be unwildly^ insensitive to local concerns, and 
generally impractical. 



A-1 



The state organizations vary widely in their approach to design. 
Some states support a central core of engineers who are responsible for the 
design of all freeway projects in the state, while others maintain resident 
engineers in various principalities to assure that new designs serve the 
local needs. In some states, the design process flows through a series 
of different offices, or bureaus, each of which controls or contributes 
to some phase of a project; the total design procedure in other states, 
from initial planning, through evaluation and approval, to final design is 
handled by a few individuals. No single procedure could possibly span 
such a wide range of practices without being accompanied by significant 
changes in organizational structure, policies, and procedures in a number 
of states. A set of guidelines, and suggestions, however, would be of 
value so that gradual changes that do occur could tend toward an optimal 
"standard" procedure. 

With regard to design selection criteria, it is probably not unfair 
CO observe that the choice of interchange configuration is more a matter 
of state custom and predisposition than of analysis. The merits and 
demerits of this situation may be argued, but given that each state organi- 
zation and procedure is largely the result of independent evolution, it 
is easy to understand how the situation has developed. 

As a consequence, a curious paradox obtains from this situation. It 

is an often quoted principle that every interchange must be designed 

in such a fashion as to marry it to the unique features of its location. 

Hong (1966) comments on this notion in his paper on interchange design: 

"Even though a set of geometric design standards has been established 
by the Bureau of Public Roads and State Highway Departments, these 
standards serve merely as a guide and not as a rigid set of rules. 
It is also true that there are many common denominators for freeway 
design features in different locations of the United States. How- 
ever, each of these freeways is unique by itself in that the 



A-2 



climatological and geographical features, and socio-economical 
characteristics may be considerably different from one region to 
another. 

"Therefore, each freeway or expressway must be custom-made to meet 
the local demands and to conform to the local topography in order 
to establish not only a functional but also an aesthetic transpor- 
tation system. It must be emphasized that the judgment and in- 
genuity of the design engineers are the main guides for judicious 
design of an efficient transportation system." 

In practice, however, a unique solution to interchange designs is not 
always applied to each site; rather, a general solution (based on past 
experience and/or predisposition) is applied and then modified to meet 
the particular requirements of the site. As a result, the state of inter- 
change design is considerably less dynamic than might be supposed. 
Further, radical (but potentially useful) approaches to interchange 
design, such as the application of set- theoretic methods proposed by 
Alexander and Manheim (1965) , have seldom found their way into the state 
procedures. 

This is not to say that interchange design is purely a matter of 
chance or bias. On the contrary, the reason that a particular design 
finds favor with a state highway department is that in the past such designs 
have met with successful operations in that state, sometimes even in the 
face of contrary logic. Thus, in Texas there are interchanges that pro- 
vide for left turns in advance of right turns — a feature that is highly 
questionable to many design engineers. In response to criticism, Texas 
officials note that the people who drive the interchanges seem to have 
no trouble with them. While Texas designers do not promote such configura- 
tions as standard, they do not hesitate to utilize them when they feel the 
circumstances warrant it. 

The^re appears to be as much disparity as conformity between practice 
and research results. Both the design texts and the research literature. 



A-3 



for Instance, stress the importance of factors such as visibility and 
uniformity of features for the convenience and comfort of the driver. 
State design personnel, however, seem to consider uniformity as being 
of secondary importance. Likewise, while some researchers identify weav- 
ing areas as major problems and recommend their elimination from freeways, 
the design texts and practioners view them as necessary, if unpleasant, 
features. 

Conversely, while most officials steer a clear path away from the 
use of left exits, research fails to provide incontestable evidence that 
left exits are, v^en properly designed and appropriate, more dangerous 
than right exits. Some studies which purport to show the hazard of left 
exits have been criticized for neglecting to correct left exit accident 
occui^i^ences for exposure. 

The following findings are also noted: 

1. Accident data are not sufficiently sensitive to the effects of 
geometries to be used as evaluative measures unless only gross geometries 
are of interest and the data are corrected for exposure. 

2. Two elements of entrance ramp terminals that are discussed at 
length by a nimber of researchers are visibility (including delineation) 
and standardization of length, taper type, and taper rate. Many authors 
suggest these factors may be more important than geometries per se , pro- 
viding that certain minimums are met. 

3. The AASHO guidelines for shape and taper rate of speed change 
lanes are generally supported by the literature, although differences 
exist between the states as to preferred shape. 

4. Both the research literature and the state design manuals are 
in accord with the AASHO guidelines for entrance ramp convergence angle. 



A-4 



5, The research data on acceleration lane length do not support the 
guidelines issued by AASHO. The use of volumes or speed differential 
between entering and through traffic as controls for length of merging 
areas are specifically criticized on the basis of logic and operational 
studies. 

6, From the standpoint of driver comfort and traffic flow, the 
circular loop appears to be superior to the elongated loop. The AASHO 
publications fail to differentiate between the two, 

7, State highway officials observe that in recent years, final 
designs are based less on optimal features and more on the least objec- 
tionable features. They are particularly troubled by the fact that local 
socio-political groups, who possess meagre information about an experience 
with roadway design can force changes in interchanges that seriously Impair 
their adequacy. 

8, Feedback from operations analysts to designers is at best poor 
and at worst nonexistent except on those Interchanges which are almost 
hopelessly Inadequate from their opening, 

9, Speed-change lanes should be designed and signed or striped in 
such a fashion as to encourage use of their full lengths. Operational 
problems on such sections are often the result of drivers using only a 
portion of the section provided and making an abrupt entry or exit, thereby 
producing turbulence in the freeway lane. 

10, Consistency in lane drop techniques is urged. Unfortunately, 
conclusive evidence as to optimal techniques has not yet been established. 
As a result, the states employ different lane drop configurations which 
are confusing to drivers unfamiliar with the particular roadway. 



A-5 



11. Ideas on and experiences with some promising new design features 
and interchange configurations fail to be promulgated outside the state of 
origin because the highway designers are too busy with daily operations 

to publish their ideas in the appropriate journals. A digest of some of 
these features and configurations will be compiled within this research 
project. It is anticipated this digest will be one of the major interest 
items in the Final Report. 

12. Many of the less-than-optimal designs and features found on the 
older interchanges may have resulted from compromises for cost reductions. 
Today, on the other hand, cost factors seem to be almost entirely absent 
from the procedures utilized in the selection and evaluation of alternative 
component configurations and in the development of design details. Until 
data are available clearly relating operational and safety benefits to 
costs of the various design features, this tendency is not likely to change. 

Major Problem Areas 

Review of the literature, current standards, and research, and dis- 
cussions of design practices with engineers in the state highway depart- 
ments reveals a number of major problem areas in the design, operation, 
and traffic control of major interchanges. These problems, listed below 
and discussed in more detail in Chapter IV, fall into three categories: 
Policy Problem Areas 

Changing Priorities 

Environmental, Aesthetic, Ecological Considerations 

Involvement with Local Agencies 

Local Access 

Partial Interchanges 

Exclusive Bus Lanes 



A-6 



Design Procedure Problems 

Uniqueness 

Design Project Management 

Selection of Basic Configuration 

Driver Needs 

Adaptability and Flexibility 

Trade-Off Analyses 

Cost Effectiveness 

Checklists 

Design Experience of Reviewers 
Component Design Problems 

Left-Hand Ramps 

Entrance Ramp Capacity 

Two -Lane Ramps 

Hidden Ramps 

Collector-Distributor Roadways 

Consecutive Ramp Arrangements and Weaving Areas 

Lane Drops 

Grades 

Signing 

Non-Conformance of Travel Paths and Construction Joints 

Freeway Traffic Control 

Nomenclature 

Traffic Forecasts 



A- 7 



Interim Report #2; Design Aids Digest 
Three-Dlmenslonal Models 

It was discovered both from the workshop and from the questionnaires 
that only occasionally are Interchange models built, and that those 
models which are built are used principally for presentation purposes 
rather than as design tools. 

As a group, the models take longer to fabricate and cost more than 
the questionnaire developers anticipated. This result obtains from the 
fact that design models (which. In comparison with presentation models, 
are primitive and lack cosmetic trim) are used much less frequently than 
was expected. 

It may be observed that engineers are generally not unfavorably 
disposed toward the use of models: on the contrary, a majority held 
that better designs resulted from their use. Their infrequent utiliza- 
tion stems rather from their low order of priority in the hierarchy of 
events which forms the design process. 

Checklists 

To the workshop participants, checklists are anathema. They are 
regarded as Impediments to imaginative thinking and as substitutes for 
professional judgment. In view of this antipathy, it was surprising 
to discover that the questionnaire respondents generally have a favorable 
attitude toward checklists, while at the same time noting that their use 
in the design process is rare. The disparity between the two groups 
may be attributable to experience. The workshops attendees were, by and 
large, seasoned veterans. It is suspected that the questionnaires were 
completed by more junior engineers, who might be more likely to depen<- or 
design aids. 



A-8 



Computer Graphics 

The general reaction of the workshop and the questionnaire popula- 
tions to computer graphics as a design aid Is one of Interest without 
enthusiasm. In both groups there prevails a posture of "wait-and-see" 
or "need more Information." In part, this attitude seems due to a skepti- 
cism regarding the costs and the efficacy of computer graphics; but 
further, there appears to be a reluctance to permit highway design 
to stray too far from human Influence toward Impersonal dictation. 

Interim Report #3; Innovative Designs Digest 
This Digest draws attention to novel Interchange designs or design 
features In the Interests of disseminating these ideas among the engineer- 
ing community for consideration in future Interchange configurations. The 
designs, pictorial and literally defined, include: 

. Turbine interchange 

. Arch-supported interchange 

. Double diamond interchange 

. Directional Interchange with left turn first 

. Major fork configuration for direct left connections 

. Anti-weave designs 

. Local access diamonds 

. Multinode interchange complex. 



A-9 



APPENDIX B 
PRE-WORKSHOP QUESTIONNAIRE AND RESPONSES 



B-i 



APPENDIX B: PRE-WORKSHOP QUESTIONNAIRE AND RESPONSES 

Introduction 

The responses to the Pre-Workshop Questionnaire froiti 12 state 
highway design engineers, A consulting engineers, 1 highway research 
engineer, and 4 FHWA engineers are tabulated in this appendix. The 
respondees are identified in Table B-1. The questionnaire consists 
largely of questions regarding different design practices and personal 
opinions regarding various design elements. Most of the questions are 
of the multiple choice type, and all invite additional comments. 

The instructions accompanying the questionnaire pointed out that 
opinion questions have no "right" or "wrong" answers, that the responses 
would be used to formulate recommendations for updating design guide- 
lines, and that specific individuals would not be identified with 
specific responses. 

A few of the questions are of the "essay type," and the responses 
are rather detailed. In large part, the answers to these questions are 
listed in summary form, deleting the repetitions; it is felt signifi- 
cant information would be lost if the answers are collapsed to tabular 
form. 

As some of the questions do not apply to some of the respondents, 
the numbers of responses cited in some of the tables do not total to 
21, Some of the tables give both the number of respondees selecting 
a particular ansWer, and the appropriate percentage. 

Major points derived from a synthesis of the questionnaire responses 
are discussed briefly in the next sub-section. These are organized by 
major topical areas, in line with the organization of the Workshop Ses- 
sions. (See Appendix C for the Workshop Agenda.) 



B-1 



Synthesis of Responses by Workshop Topics 
Standardization (Questions: 8, 9, 17) 

The respondents felt that standardizing some aspects of the geo- 
metric design of major Interchanges Is possible, but generally not 
practical. Of those who do not feel that It is even possible in urban 
areas, most remarked that the unique conditions at each site are the 
major deterrents to standardization. For rural areas, topography 
appears to be a dominant factor in design decisions. Restrictions of 
cost, space availability, and special traffic conditions are the 
dominant arguments against the practicality of standardized urban 
designs. 

Design Process (Questions: 1, 2, 3, 46) 

It appears that the weighting assigned to the various factors 
which design engineers use to determine geometries of a major inter- 
change is variable. Two respondents state that the factors cannot be 
listed in any particular order. The design volumes and level-of-service 
appear to be the more important factors. The respondents also feel that 
the major differences between major interchanges and others are pro- 
vision for continuous movement and basically a higher level of service. 
Finally, more than two- thirds of the respondents feel that it is desir- 
able to have a separate design procedure for major interchanges. 

Configuration Evolution (Questions: 1, 2, 27) 

Question 27 is the only one dealing specifically with overall 
configurations, although Questions 1 and 2 are relevant. Lower cost 
and acceptability with lower turning volumes are cited as factors 
favoring the clover leaf. 



B-2 



Cost-Effectiveness , Trade-Of f s (Question: 19) 

The number of engineers who will accept a left-side ramp increases 
as the relative cost savings of the left side ramp over one on the 
right increases from $100-500 thousand. One respondent indicated the 
savings would have to exceed $1 million before he would seriously 
consider using a left-side ramp in a major interchange,, 

Visibility, Design Aids (None) 

Design aids are discussed in Interim Report 2. 

Exits (Questions: 10-24, 34, 37, 38) 

The respondents generally feel that the single exit for both turn- 
ing movements is the most desirable, even for exits requiring two 
lanes. The two-exit configuration, with the right turn taken off first, 
is the next choice. Left-hand exits are almost never usedr 

The opinions on the desirable lengths between exit gores varied, 
and no one value dominated with a high percentage of selection. 

Tapered deceleration lanes are preferred and used in most situa- 
tions. The parallel type is used for situations where ramp and mainline 
speed differences are too great, sight distance is restricted, the 
mainline is on curve, or there is a possibility of back-up onto the 
through lanes. 

Entrances (Questions: 12-1-17, 25, 26, 28-33, 39-43) 

Left side entrances are generally deemed permissible only when there 
is no space available for a right-side entrance, or cost of the alter- 
natives is prohibitive, and the entrance lane(s) is added to the main- 
line. A single entrance (two ramps form one ramp prior to entrance 
terminal) to the mainline is almost unanimously deemed preferable over a 



B-3 



double entrance. Left entrances are rarely used, and the single 
entrance from the right shows a slight edge of use over the double 
entrance (both on the right) o Again, if a lane is added, there is 
less objection to the left-side entrance. 

Nearly half the experts would use the double entrance with an 
adequate separation distance rather than a single entrance requiring 
two lanes. Again, the values for that distance are almost evenly 
distributed in most cases. 

The taper type acceleration lane for entrance ramps is generally 
preferred, but the parallel is used in some situations, such as 
restricted length, very heavy mainline flow, or poor sight distance, 
A 50:1 taper is felt to be the most desirable value for the single 
entrance. Fifty-seven percent of the experts felt that the right lane 
should be dropped for that situation. 

On the mainline, the merge of the two entrance lanes shows dominance 
of use over the merge of the left entrance lane and the right main- 
line lane,, Finally, the merge C'f the two right lanes showed clear 
dominance over merges of either the Cssnter or right lanes » 

Route Continuity; Ramp Arrangements (Questions: 12-17, 19-24, 26, 28, 

30-32, 44) 

The collector-distributor road for two adjacent loop ramps is used 
more in urban areas than in rural areas j and more often than a design 
with no C-D road. The responses regarding the desirable distance 
between the loop ramp terminals do not make any one value the clear- 
cut choice r^ 

About 80 percent of the respondents feel that weaving areas should 
always be avoided in major incerciianges, yet 70 percent state that 



B-4 



i 



weaving areas can be justified with adequate weaving lengths Also, 
75 percent feel that weaving sections are acceptable if they are off 
the through roadway. 

Lane Drops; Lane Balance (Questions: 35, 36, 40-43) 

Almost no one uses a left-lane drop adjacent to a right-side exit 
terminal. For the two-lane exit from a basic 8-lane freeway, the right 
side lane drop adjacent to the exit terminal shows clear dominance over 
the design where the lane is carried through the interchange. 

The respondents clearly prefer the right side mainline lane drop. 
Also, the right lane is chosen as the lane to be dropped where a two- 
lane turning roadway is merged into one lane before the entrance ter- 
minal at the mainline » For the merge of two 2-lane roadways into one 
3-lane roadway, the design where the two right lanes are merged is 
clearly preferred over the alternatives. 

Local Access; Freeway Control (Questions s 4-7) 

The respondents indicate that political pressure sometimes dictates 
the provision of local ramps in a major interchange, but that this is 
not usually the case. Two-stage public hearings have some impact on 
local ramp provision. Practically all the respondents feel that local 
access should not be permitted in a major interchange- 

Opinion is divided on whether selective closure of ramps during 
peak periods is a practical solution to capacity-operational problems. 



B-5 



TABLE B-1. Respondents to Pre-Workshop Questionnaire 



Design 

Churchill, Robert R. 

Deputy Design Engineer for Roadways 

Florida Department of Transportation 

Dayton, Edwin W. 

Chief, Bureau of Surface Design 

New Jersey Department of Transportation 

Everhart, B. F. 

Chief Design Engineer 

Ohio Department of Highways 

Foster, W. M. 

Assistant Director of Highway Development 

Washington State Highway Commission 

Foy, Robert A. 

Chief Engineer 

Design Division 

Wilbur Smith & Associates 

Gazda, Andrew J. 

Engineer of Geometric Design 

Illinois Department of Transportation 

Hall, Parker L. 

Assistant Engineer for Design 

California Division of Highways 

Hlbbs, John 0. 

Regional Design Engineer, Region 3 

Federal Highway Administration 

Hofmann, Frederick J. 
Senior Highway Engineer 
Edwards and Kelcey, Inc. 

Housworth, Jack L. 
Supervising Design Engineer 
Texas Highway Department 

Kenyon, Alan D. 

Associate Civil Engineer 

New York State Department of Transportation 



B-6 



Lins, William F. 

Chief, Bureau of Highway Design 

Maryland Department of Transportation 

Loutzenheiser, Donald W. 
Chief Highway Engineer 
Federal Highway Administration 

McCoy, William D. 

Assistant State Highway Urban Engineer 

Georgia Department of Transportation 

Mueser, Robert R. 

Deputy Chief Highway Engineer 

Pennsylvania Department of Transportation 

Pennington, Gordon R. 

Sverdrup & Parcel Associates, Inc. 

Randich, G. M. 
Vice President 
Deleuw, Gather & Company 

Sigal, Andre H. 

Associate Civil Engineer 

New York State Department of Transportation 

Region 10 Office 



Operations 

Taragin, A. 

Traffic Performance and Analysis Division 

Office of Traffic Operations 

Federal Highway Administration 



Academic - Research 

Glennon, John C. 

Midwest Research Institute 

Pilkington, George 

Federal Highway Administration 

Environmental Design and Control Division 



B-7 



1. Please list in order of importance the most critical factors used in 
determining the basic geometries of a major interchange. 

Factors: 

a. Traffic Mainline Volumes (Design Hour) 

b. Site conditions - environment, topography 

c. Standards and capacity of crossing freeways (Level of Service; 
use of direct vs. loop, left vs. right on-off) 

d. Economics 

e. Project Objectives 

f. Weaving distances, grades, acceleration-deceleration lanes 

g. Simplicity of design 
h. Urban vs. rural 

i. Right-of-way 

j . Route and lane continuity and lane balance (freeway turns) 

k. Design speed 

1. Sight distance 

m. Spacing between adjacent interchanges 

n. Safety 

o. Operational characteristics 

p. Ability to obtain community approval (political constraints) 

q . Signing 

r. Effect abutting properties (socio-economic effects) 

s. Appearance 

t. Interchanging traffic volumes 

u. Ramp design speeds 

V. Lane drops on right 

w. Land uses and development 

X. Angle of intersection 

y. Construction controls (funding, schedules, maintenance of traffic) 

z. Alignment of approaches to points of divergence, 

za. Percent of heavy trucks (composition of traffic) . 

Note: Factors were provided by experts; therefore number of factors 
varied for each individual. 



B-8 



Number of Experts (out of 21) Ranking Factors 

RANK 
23456789 



10 



11 



12 





a 


6 


4 


4 










b 


1 


2 


4 


1 


1 






c 


5 


3 


1 


2 


1 






d 






3 


3 








e 






1 










f 








1 








g 








1 








h 


1 


1 












i 


2 


1 


2 


2 








J 


1 


2 




1 








k 


1 


3 




1 








1 




1 










w 


m 










2 




o 


n 


2 


1 


1 








H 


o 






2 






1 


CJ 
















<: 


P 












1 


p^ 


q 

r 
s 






1 


1 
1 




1 
2 




t 


1 


1 ' 






1 






u 




1 












V 






i 










w 




1 












X 


1 














y 










1 




' 


z 














1 


za 






1 


1 







Note: While some categories might be combined, the actual wording of the 

answers suggest significant or at the least, discernible differences 
between most of the categories. 



B-9 



D, 



One of the principal goals of this major interchange study is to 
determine the differences in the design approach and design proce- 
dures for freeway-to-freeway interchanges compared to other inter- 
changes (i.e., freeway-to-arterial, expressway-to-local highway, freeway- 
to city street, etc.) Please state your opinion regarding the major 
differences between design of a major interchange and other (non-m.ajor) 
interchanges. Please list the differences in order of importance. 



1. Major difference is that left turns (intersections) are eliminated 
at freeway-to-freeway interchanges. 

2. Major interchanges must accommodate higher percentage of unfamiliar 
drivers - greater driver expectation for higher standards. 

3. Major interchange design should allow more flexibility for change 
due to traffic projection inaccuracies. 

4. In urban areas, major interchanges should be in balance with the 
entire freeway network - muat look at total picture. 

5. Major interchange design is more complex, more costly, has greater 
impact on the community and the freeway system than does a local 
Interchange. A greater effort is necessary to overcome greater pro- 
blems . 



CO 

<: 

M 2. Design speed 



1. Traffic volumes 



3. Level of service 



> 

Q 4. Safety features, i.e., width of clear area, rate of slopes 



5. Impact attenuators 



1. This, as above, cannot be given a particular number of importance. 
A chain is no stronger than its weakest link; therefore, if a major 
interchange on a particular facility is the ultimate in design fea- 
tures and the freeway-to-city streets is poor design; the facility's 
general operation is poor due to the freeway-to-city street "bottle- 
neck." Every l/C on a facility or a system should be designed to 
insure that the overall system is sufficient. 



1. Higher ramp speeds must be maintained. 

2. Initial consideration should be for minimum two lane direct 
connections. 

3. Interchange locations, must give consideration of additional 
right-of-way requirements. 



B-10 



E. 



F. 



G. 



M. 



1. Major Interchange must provide continuous movements for all 
strands of traffic. 

2. All turning roadways of a major Interchange must provide minimum 
reduction in design speed from that of mainline roadways. 

3. Directional design of major interchanges may result in left side 
terminals. There should be none in other interchanges, 

4. There may be a reduction in number of mainline lanes through a 
major interchange. There should be no lane drops through other 
interchanges. 

5. Major interchanges often require C-D roads or auxiliary lanes 
to facilitate weaving and turning movements. 



1. Higher level of service 

2. Higher speed design 

3. Operational characteristics can be less strict for a non-major 
interchange (merging, weaving, direction of movement, signing). 

4. Local service to existing streets should be goal of minor inter- 
change. 

5. Different configuration of major interchanges and minor inter- 
changes depending upon type of traffic (i.e., thru or local traffic) 



1. Major interchanges usually require direct connections to provide a 
high level of service. Minimum reduction in freeway design speed 
desirable. 

2. Minor connections should be eliminated, such as service to local 
roadways and U-turn facilities. 

3. Weaving areas - undesirable. 

4. Major volumes should be designed as a through traffic move pre- 
ferably with the minor moves using right hand off. 



1. Level of service - lower levels of service are tolerated on non- 
major interchanges before higher type facilities (directional ramps, 
C-D roads) are added. 

2. Minimum interchange type - with (4-leg) major interchanges, a clover- 
leaf is the assumed minimum type before item (1) steps (a) to (g) 
are applied to determine maximum needs. For all other cases, a dia- 
mond type is used as starting point. 

Note: item (1) etc. refers to answer for Question 1. 



B-11 



N. 

1. Characteristics of the traffic. That is, a major interchange 
handles traffic from 2 high speed freeway type facilities. Lesser 
interchanges handle traffic entering or leaving a roadway where 
traffic speed may be lower, interruptions to traffic may be frequent, 
at grade intersections exist, etc. 

2. With a major interchange, free-flowing traffic through the inter- 
change ramp is desirable. With lesser interchanges, ramp terminals 
at the crossroads may be other than free-flow. 

3. Major interchanges direct attention to through traffic. Lesser 
interchanges should consider local traffic demands to a far greater 
extent . 

0. 

1. Higher interchanging volumes. 

2. All ramp terminals must be of directional type (no traffic conflicts 
can be tolerated) . 

3. Design speeds are higher on both mainlines and therefore ramp 
design speeds should be higher. 

4. Major interchanges generally involve multiple structures and/or 
multi-level structures. 

P. 

1. All maneuvers are free-flowing on major interchange, not usually 
required on lesser interchanges. 

2. Higher speeds are normally accommodated in major interchanges and 
the operational integrity of both freeway facilities is of major 
importance whereas on less significant interchanges, nominal sacrifice 
on the lesser roadway may be justified. 

3. Lane balance and continuity are of extreme importance on major 
interchanges. 

4. No surprises — this is especially true with higher operating speed - 
major interchanges. 

5. Excellent signing required — the more complex the interchange and 
higher the operating speeds the more critical this becomes. 

6. Compound weaving must be avoided in major interchanges. This is 
undesirable on all interchanges, however, it may be tolerated on 
less significant interchanges with lower volumes and lower operating 
speeds. 

R. 

1. Design approach — same for all interchanges. 



B-12 



i 



Research 

A. No response 



M. 

1. Higher turning volumes, 

2. Desire to maintain level of service, 



FHWA 

A. No response 

B. 

1. Basic design is not really different or the procedures used. 

2. Major has higher speed, free-flowing ramps as opposed to diamond 
ramps or loops. This entails greater lengths, added lanes, longer 
terminals, larger sight distance, etc. that tends to approach 
through traffic operational layouts. May be C-D roads to separate 
weaving and frequent access . 

3. Spatially large and a greater effect on the site area. Layout 
results in sizeable open areas, extracted from valuable development 
potential land which must be made into attractive, fitting open 
space and efforts made to include appropriate joint development 

or other community aiding uses. 

4. Fceeways are alike so design for two does not have to be compromised 
to fit the limitations and operational restraints of the normal 
crossroad, often existing, of a lower type. 

C. 

1. Major necessitates a free-flow type ramp, therefore many inter- 
change types are eliminated. The choice is basically a full clover- 
leaf, a directional, or combination thereof. 

2. Signing is more critical for major Interchange designs because 
more or most drivers must make decisions at major forks. 

3. Alignment or view of road approaching major decision points is a 
prime item for consideration in design of major forks (no sign is a 
substitute for being able to see the roadway pavement ahead) . 

4. Traffic - necessary to determine the type ramps and number of lanes. 
Less weight should be given to traffic as compared to achieving 

good "lane balance" at ramp terminals and "lane continuity." 



B-13 



Consultants 
A. 



B. 



M. 



N. 



1. For major Interchanges, all traffic should be considered as through 
moving traffic regardless of direction of movements within interchange. 

2. All traffic should be considered as operating under high speed 
conditions without interruption. 

3. No local access or ramps should be provided. 

4. Greater emphasis must be placed on traffic flow continuity and less 
on sheer capacity. 

5. Weaving, merging and stop conditions should be avoided. 



1. High speed geometries. 

2. Greater safety demands dictated by the higher speeds. 

3. Direct ramps (not loops). 

4. No local access to streets/roads. 

1. Design speed required on turning roadways and terminals. 

2. Larger volumes of traffic to accommodate (2-lane ramps, perhaps) 
(direct connections) . 

3. Maintenance of existing traffic. 

4. Lighting, signing. 

1. Recognition route continuity characteristics. 

2. Design for higher speeds and operating conditions. 

3. Complexity. 

4. Adaptability of design to accept signing. 

5. Spatial requirements. 

6. Dollars invested. 



B-14 



p. 



R. 



1. Provision of directional connectors, 

2. Elimination of "surprise" elements. Design should conform to 
driver expectancy. 

3. Design speed on ramps should be within 20 mph of the approach speed, 



1. Facility should be capable of operating at or close to the operat- 
ing speeds of the through roadways. 

2. Facility should be capable of handling the volume and composition 
of traffic both in the through and turning lanes. 

3. Facility should "blend" with the area; i.e., it should not present 
the driver with any conditions he would not expect to find in the 
area. 

4. Facility should maximize safety demands. 

5. Should demand more "fore-thought" and sensitivity to non-highway 
demands than has been exercised in the past. 



B-15 



Using as a reference the Intersection Design Procedure as summarized in 
the AASHO "Blue Book" (pages 603-4), do you feel the design of a major 
interchange is sufficiently different from the design of other (non-major) 
freeway interchanges to make a separate design procedure for major inter- 
changes either necessary or desirable? 



Choices 

a. There is a definite necessity to 
have a separate procedure for the 
design of major interchanges 



Number of Participants 
Selecting Given Answer 



23.8 



The same procedure can be used 
for both types of interchanges 
but it would be desirable to have 
a separate procedure for major 
interchanges 

The same procedure should be used 
for major and non-major inter- 
changes 

Total Number Responding 



21 



33.3 



42.9 



100 



Comments : 

Major interchanges should continue to be "custom" designed while 
the use of uniform designs (more diamonds) should be the objective for 
non-maj ors . 

The same procedure should be used with different standards. 

The same procedure should be used although the study of major inter- 
changes must be in greater depth. Some steps may be eliminated or 
curtailed for non-major interchanges. 

The same procedure should be used but the extent and details differ. 



B-16 



4. In a state funded project, how often do political or public pressures 
in your state dictate that local ramps be designed into a major 
interchange? 



Number 
Responding 

Percent 



Almost never Sometimes Often Usually Almost always Total Number 
(0-5%) (6-35%) (36-65%) (66-95%) (96-100%) Responding 



4 
22.1 



11 

61.1 



1 
5.6 



1 
5.6 



1 
5.6 



18 
100 



Comments : 

I think that practically all designers know better, it is just that 
we do not know how to resist the pressures. 



5. In a federal/state funded project requiring the two stage public 
hearings, how often do such hearings result in the requirement to 
design local ramps into a major interchange? 

Almost never Sometimes Often Usually Almost always Total Number 
(0-5%) (6-35%) (36-65%) (66-95%) (96-100%) Responding 



Number 
Responding 

Percent 



27.8 



11 



61.1 



11.1 







18 



100 



Comments : 

Our public hearings requirements are the same for 100% state as 
50-50 projects on Interstate. 

Most pressures are brought before the public hearing. The hearings 
are often a formality. 

I have not experienced such requirements because of limited public 
hearing involvement regarding major interchanges. 



B-17 



6. From a design and operations standpoint, how often do you think local 
access ramps should be permitted in a major interchange? 



Almost never Sometimes Often Usually Almost always Total Number 
(0-5%) (6-35%) (36-65%) (66-95%) (96-100%) Responding 

19 

100 



Number -g 
Responding 


3 











Percent 84.2 


15.8 











Comments: 











This is dependent on location and area needs. If needs can be satisfied 
within reasonable distance, then no local connections should be permitted. 

Confusion and erratic operations often result from local ramps in major 
interchanges due to signing difficulties. Where proper signing can be 
provided a local ramp would be acceptable, but this is rarely possible. 

Sometimes, but they must not reduce the ef f ectiveness cf interchange 
movements . 

Local access should be planned for and gained from properly spaced 
interchange with local roads . 

Costs and signing, etc. complications indicate we should avoid mixing 
freeway-to-freeway interchanges with local interchanges. This is not 
always reasonable in urban situations and there are acceptable way to 
combine the two types. 

I think that practically all designers know better, it is that we do 
not know how to resist the pressures. 

This should be based on traffic desire flow and not a "fixed" figure. 
You may recall that the old BPR had a fixed number of ramps and distance 
spacing between interchanges which created "bottlenecks" at the few ramps 
provided and created a "Chinese Wall" through urban areas. 

Occasionally special geometric conditions may prevail where opera- 
tions would not be hampered by the introduction of a local ramp connection. 
A local ramp connection to a low volume interchange would be a case in 
point 

Local access ramps always create operational and signing problems when 
incorporated into major interchanges. They cause driver confusion and there- 
fore safety problems. Local ramps should thus rarely be incorporated into 
a major interchange and should be avoided within the influence zone (1 mile +) 
of a major interchange. 

Problems have arisen when such is permitted. European practice bears this out 

Local access should only be permitted in major interchanges where there 
is no other way to feasibly allow access, and projected local access traffic 
volumes would not force the facility to operate below level of service "A". 

B-18 



7. In some Instances local ramps located within a major interchange are 
selectively closed during peak periods. In your opinion, is selective 
closure of ramps during peak periods a practical solution to capacity- 
related operational problems which may exist? 



Number 
Responding 



Almost never Sometimes Often Usually Almost always Total Number 
(0-5%) (6-35%) (36-65%) (66-95%) (96-100%) Responding 

5 4 6 3 2 20 



Percent 



25.0 



20.0 



30.0 



15.0 



10.0 



100 



Comments : 

This is not recommended — more satisfying and permanent solutions should 
be sought. 

This depends on alternate connections or facilities within the corri- 
dor and whether traffic could be adequately redirected. 

If such is anticipated prior to construction, the ramps should not be 
built. 

Selective closure of ramps is practical only if acceptable alternate 
routes are available during the period of closure. The community must be 
ready to accept this type of restraint. 

Selective closure of ramps during peak periods is an effective solu- 
tion for existing conditions, but generally should not be designed for 
that condition on new freeway designs. 

Since it assiimed the problem exists (probably from poor planning) , 
selective closure often is a practical remedy in large metropolitan areas. 

Almost never — metering should be considered. 

We should use this experience to resist building more local service 
ramps too near major interchanges. 

A better solution would be to eliminate the ramp entirely and require 
some vehicles to travel greater distances on local roads. Local opposition 
sometimes makes this solution impractical. 

Local ramps are not closed during peak periods in any interchanges in 
the state of Maryland. There are only two local off-ramps in Maryland. 
No need to close in peak hour. 

The ramp should be removed permanently if practical, without severe 
damage to local businesses. 



B-19 



This will help keep the traffic flow of the mainline at an acceptable 
level. Local traffic conversely will be more congested. 

If a ramp is not usable during peak hours it has questionable value 
in the system. 

Basic objective of a major interchange is to effect a freeway-to- 
freeway movement. This function should be safeguarded particularly the 
main roadway Is severely congested and local ramps are hampering traffic 
operations. 

Once a ramp is constructed and opened to traffic, it should not be 
intermittently closed. This type of closure causes driver confusion and 
related safety problems. This is especially true In the case of local 
exit ramps, more latitude is possible with entrances. 

This is only a stop-gap solution. 

A better method would be to close them permanently. 



B-20 



8. 



In reviewing the literature relevant to major interchange design and 
operations, it was found that a number of authors proposed some stand- 
ardization of geometric design so that the driver is faced with a more 
constant exiting/entry situation. In your opinion is the standardiza- 
tion of geometric design of major interchanges feasible and practical 
for the situations listed below? 



Urban Interchange 







Feasible? 






No. 


Responding 


Percent 


Yes 




12 


63.2 


No 




7 


36.8 


Total 




19 


100.0 



Rural Interchange 







Feasible? 






No, 


Responding 


Percent 


Yes 




15 


78.9 


No 




4 


21.1 


Total 




19 


100.0 



Practical? 
No. Responding Percent 
8 38.1 

13 61.9 

21 100.0 

Practical? 
No. Responding Percent 
13 61.9 

8 38.1 

21 100.0 



Comments : 

Standardization, yes; uniformity, no (all "no" responses above). 

It is related to the siting situation (all "yes" responses). 

I assume we are talking about terminal configurations and not the actual 
shape of the ramps (all "yes" responses). 

The word some is underlined. Standardization or uniformity should rate 
high in the priority of design considerations, but a "cookbook" approach is 
not feasible, especially for major interchanges. 



B-21 



9. If you answered "no" to any part of question //8, please list, in order of 
importance, those factors which argue against the feasibility or practical- 
ity of standardizing the geometric design of major interchanges. In list- 
ing the factors, please indicate the relevant situation as follows: UF » 
Urban Feasibility; UP = Urban Practicality; RF = Rural Feasibility; RP - 
Rural Practicality. 

Urban Feasibility : 

When thinking of standard design what comes to mind is the single exit, 
single entrance design connected to collector-distributor roads. This is a 
desirable thing to do but is not always feasible in heavily built up urban 
areas because of the added right-of-way required. 

Usually a limited area is available and interchanges are tailor-made 
to fit existing conditions. The economical aspect is probably the biggest 
deterrent in standardization. 

The proximity of exit and entrance terminals in large metropolitan areas 
often makes it infeaslble to shift certain terminals just for uniformity. 
Also the need for more multi-lane directional ramps (major forks) in these 
areas and ramp splits reduces the feasibility in these areas. 

In urban areas factors other than a standardized configuration assume 
such magnitude that it is necessary to design each interchange to fit the 
particular circumstances. 

Proper signing negates the need for standardization since strangers 
depend entirely on signing, not geometries. There are also terrain, topo- 
graphical, and right-of-way restrictions. 

Right-of-way attainability 
Environmental and social effects 
Influence of adjacent interchanges 
Time lapse in system development. 

Urban Practicality : 

Right-of-way restrictions 
Economics 

Due to restrictions of development and right-of-way costs and variations 
In traffic needs and street patterns, the complete standardization of urban 
Interchanges is not practical. 

Many times in urban areas the land is simply not available to standardize 
design. 

Usually a limited area is available and Interchanges are tailor-made to 
fit existing conditions. The economical aspect is probably the biggest deterrent 
In standardization. 

Our studies (111.) have shown that alternative designs to accomplish uni- 
form exits and entrance have all Involved additional C-D roads, greater struc- 
ture costs, more right-of-way, etc., to the point where it was not considered 
the optimum design. 

B-22 



In urban areas factors other than a standardized configuration assume 
such magnitude that It is necessary to design each interchange to fit the 
particular circumstances. 

Right-of-way costs involved in the more desirable concept are sometimes 
so excessive that compromise is necessary. Special traffic conditions may 
on selected occasions indicate the desirability to construct non-standard 
configurations. Such situations are isolated in occurrence (certain non- 
standard configurations may eliminate or minimize weaving, etc. — depending 
upon the proximity of other entrances and exits) . 

Economics, due to dense development in urban areas, makes standardization 
unrealistic. 

Proper signing negates the need for standardization since strangers 
depend entirely on signing, not geometries. Standardization would result 
in higher right-of-way and higher construction costs. 

Right-of-way constraints 
Physical restrictions 
Cost 

Rifeht-of-way attainability 
Environmental and social effects 
Influence of adjacent interchanges 
Time lapse in system development 

Rural Feasibility ; 

If all interchanges required the same type of connections and were at a 
similar crossing angle, it might be possible. But topographical features 
sometimes dictate changes in design from a more desirable type interchange. 

Proper signing negates the need for standardization since strangers depend 
on signing, not geometries. There are also terrain, topographical, and right- 
of-way restrictions. 

Right-of-way attainability 
Environmental and social effects 
Influence of adjacent interchanges 
Time lapse in system development. 

Rural Practieality : 

Right-of-way restrictions 
Economics 

If all interchanges required the same type of connections and were at a 
similar crossing angle, it might be possible. But topographical features 
sometimes dictate changes in design from a more desirable type interchange. 

Our studies (111.) have shown that alternative designs to accomplish 
uniform exifefi^and entrances have all involved additional C-D roads, greater 
structure costs, more righteef-way, etc., to the point where it was not 
considered the optimum design. 

B-23 



while it may be feasible to standardize geometric design for rural inter- 
changes because of fewer restrictions due to existing land use, it is not 
practical. A better design, and often a more .economical one, will result 
from tailoring the design to the circumstances. 

Proper signing negates the need for standardization since strangers 
depend entirely on signing, not geometries. Standardization would result 
in higher right-of-way costs and higher construction costs. 

Right-of-way attainability 
Environmental and social effects 
Influence of adjacent Interchanges 
Time lapse in system development 

General Comments ; 

Parts of any freeway-to-freeway interchange can and should be standardized 
(exiting and entering, etc.). I doubt benefit to any "standardization" of 
the interchange as a whole — no two situations are ever the same. A good 
line must be drawn between standardization for consistency which is desirable 
and standardization that will stifle "thinking." 

It should be noted that while entire Interchanges should not be standardized, 
the various components that make up an interchange are normally standardized, 
thus maintaining consistency in design. 



B-24 



10. In your organization how frequently are loop ramps used in new (i.e., 
recently designed or constructed) freeway-to-freeway interchanges for 
major turning movements? 



Number 
Responding 

Percent 



Almost always Usually Often Sometimes 
(96-100%) (66-95%) (36-65%) (6-35%) 



1 
5.5 



Almost never 
(0-5%) 



14 

77.8 



14.7 



Total Number 
Responding 



18 



100 



Comments : 

Almost always where single lane capacity is not exceeded and the 
route does not turn. 



11. In your organization how frequently are loop ramps used in nev^ freeway- 
to-freeway interchanges for minor turning movements? 



Number 
Responding 

Percent 



Almost always Usually Often Sometimes Almost never Total Number 
(96-100%) (66-95%) (36-65%) (6-35%) (0-5%) Responding 



5.6 



33.3 



33.3 



22.2 



5.6 



18 



100 



Comments : 

Almost always where single lane capacity is not exceeded and the route 
does not turn. 

Almost never would be a preference if given such a choice. Combinations 
of loops with diamond ramps in some quandrants are even a lower choice, 
except for possibly the Parclo "A", 4-quandrant design where left turn desire 
along the crossroad is so heavy that signalization is not possible. 



B-25 



12. Figure 1 in the "Figures Package" shows 2 adjacent loop ramps, without a 
collector-distributor road, and Figure 2 shows the same arrangement, with 
a collector-distributor road. Place an "X" under the use category, which 
characterizes the frequency with which your organization uses these 
designs for freeway-to-freeway Interchanges, under the following conditions: 






USE CATEGORY 



CONDITION 



a. Fig. 1 Rural Areas No. Responding 

% 

b. Fig. 1 Suburban Areas No. Responding 



c. Fig. 1 Urban Areas 

d. Fig. 1 Max. Vol. — 
each loop ramp < 100 vph 

e. Fig. 1 Max. Vol. ~ 

each loop ramp 100-300 vph 

f. Fig. 1 Max. Vol. ~ each 
loop ramp 300-500 vph 

g. Fig. 2 Rural Areas 



No. Responding 
% 



00 

to 

^^ 

< B-2 
O 
4J O 
CO i-l 
O I 

e ^ 



2 
11.8 




1 

6.3 

5 
31.3 

1 
6.7 

1 
5.9 

2 

11.8 



CO vO 



1 
5.9 

3 
17.7 




2 
12.5 

5 
33.2 



0) I 





V4 




V4 






0) 




(U 






> 




.q 


bu 


(D 


0) 




tl 


C 


0) 


fz 




3 


•H 


^ 






;z: 


TJ 


•H ^s 


■M 


^-s 




C 


■u in 


CO 


6^ 


i-i 


o 


<U CO 


O 


m 


to 


p- 


6 1 


a 


r 


•u 


co 


o \o 


rH 


o 


o 


"i 


w ^-^ 


O 


>wX 


H 


Pi 



4 
23.5 

3 
17.7 

3 
18.7 

1 
6.2 

1 

6.7 



2 4 
11.8 23.5 



4 
23.5 



7 
41.1 

4 
23.5 

3 
18.7 

4 
25.0 

4 
26.7 

3 
17.7 

7 
41.2 



3 
17.7 

7 
41.1 

9 
56.3 

4 
25.0 

4 

26.7 

7 
41.1 

4 
23.5 



17 
100 

17 
100 

16 

100 

16 
100 

15 
100 

17 
100 

17 

100 



B-26 



USE CATEGORY 



CONDITION 



CO 

>. 

CO 
& 

O 

■u o 

CO iH 
O I 

e ^ 



n] I 



OJ 1 

-U VO 

U-^ CO 

O ^-' 



en 

0) 

e /-^ 
•H B^ 

•U LTl 

0) ro 

e 1 

o ^ 
CO ■^-' 



J-l 

(1) 






O 

e I 

iH O 



o 00 

e c 
c 

iH O 
O QJ 



h. Fig. 2 Suburban Areas No, 

i. Fig. 2 Max. Vol. — 
each loop ramp 
< 100 vph 

j. Fig. 2 Max. Vol. — each 
loop ramp < 100 vph 

k. Fig. 2 Max. Vol. — each 
loop 100-300 vph 

1. Fig. 2 Max. Vol. — each 
loop ramp 300-500 vph 



Responses 
% 



1 
5.9 

1 
6.3 

1 
6,2 






3 
17.6 

3 

18.7 




2 
13.3 



3 
17.6 

7 

43.8 




1 
6.7 



9 
53.0 

2 

12.5 

5 
31.3 

8 
53.3 

6 



1 
5.9 

3 

18.7 

10 
62.5 

4 
26.7 



17 
100 

16 

100 

16 
100 

15 
100 

17 



11.8 11, 



29.3 35.3 11.8 100 



Comments: 

Our (Illinois) use of Figure 2 in rural areas would generally be for weaving 
capacity needs, whereas in urban areas it would generally result from a preplanned 
C-D road to collect closely spaced ramps. However, without preplanned C-D, it 
is also used for capacity more often in urban than rural areas. 

In Maryland directional ramps are used in urban areas. 

The above represents past practice in our organization (New York) . It is 
assumed that, with publication of our new design manual, the use of configuration 
1 will virtually disappear for freeway-to-freeway interchanges. 



B-27 



13. Indicate the minimum and desirable distance D between entrance and exit 
nose for figures 1 and 2 under the following conditions. The answer 
is to reflect your opinion as a designer and not necessarily the values 
presented in your state design manual, blue-book, red-book, etc. 



\~ 




r 



D 





Fig. 1 Mainline Design Speed 50 mph 



Minimum Distance D (ft.) 

400 

450 

500 

600 

700 

800 

900 
1000 
1400 
1500 
1700 



Number Respondin 


g Percent 


Cumulative % 


1 


5.0 


5.0 


1 


5.0 


10.0 


3 


15.0 


25.0 


1 


5.0 


30.0 


3 


15.0 


45.0 


4 


20.0 


65.0 


2 


10.0 


75.0 


2 


10.0 


85.0 


1 


5.0 


90.0 


1 


5.0 


95.0 


1 


5.0 


100.0 



Total 



20 



100.0 



Comments : 

A 4-quadrant cloverleaf is obviously not practical for these distances 
(1400 min., 2000 desirable). Today at 4-quadrant cloverleafs we would pro- 
vide C-D roads. 

700 ft., subject to weaving criteria (quality of flow II for fig. 1 and 
III rural or IV urban for fig. 2). 



B-28 



Desirable Distance D (ft.) 

600 

700 

800 

900 
1000 
1200 
1300 
1500 
2000 



Number Respondii 


ig Percent 


Cumulative % 


1 


5.9 


5.9 


1 


5.9 


11.8 


2 


11.7 


23.5 


1 


5.9 


29.4 


5 


29.4 


58.8 


2 


11.7 


70.5 


1 


5.9 


76.4 


1 


5.9 


82.3 


3 


17.7 


100.0 



17 



100.0 



Total 

Comments: 

The distance Is based on weaving volumes (no specific answer given) . 

A 4-quadrant cloverleaf is obviously not practical for these distances 
(1400 min., 2000 desirable). Today at 4-quandrant dloverleafs we would pro- 
vide C-D roads. 

Should not exist. 



Fig. 1 Mainline Design Speed 70 mph 
Minimum Distance D (ft.) Number Responding Percent 



450 

500 

700 

800 

1000 

1200 

1300 

1400 

1800 

2000 

2300 

Total 

Comments : 



1 
2 
1 

1 
4 
3 
2 

1 
1 
2 

1 

19 



5 


.3 


10 


.5 


5 


.3 


5 


3 


21 





15 


7 


10 


.5 


5 


3 


5 


3 


10 


5 


5 


3 



100.0 



Cumulative % 

5.3 
15.8 
21.1 
26.4 
47.4 
63.1 
73.6 
78.9 
84.2 
94.7 
100.0 



Distances greater than 800-1000 are beyond the realm of practical loop 
design. 

A 4-quadrant cloverleaf is obviously not practical for these distances 
(1400 min., 2000 desirable). Today at 4-quadrant cloverleaf s we would pro- 
vide C-D roads. 



B-29 



Number 


Resp( 


anding 


Percent 


Cumulative % 


1 






5.9 


5.9 


4 






23.5 


29. A 


1 






5.9 


35.3 


1 






5.9 


41.2 


3 






17.6 


58.8 


1 






5.9 


64.7 


1 






5.9 


70.6 


3 






17.6 


88.2 


1 






5.9 


94.1 


1 






5.9 


100.0 



Desirable Distance D (ft.) 

900 
1000 
1200 
1400 
1500 
1600 
1700 
2000 
2600 
3000 

Total 17 100.0 

Comments: 

1000 ft. would require spreading the crossroad lanes or very large 
radii on the loops. 

The distance is based on weaving volumes (no specific answer given) . 

A 4-quadrant cloverleaf is obviously not practical for these distances 
(1400 min., 2000 desirable). Today at 4-quadrant cloverleaf s we would 
provide C-D roads. 

Distances greater than 800-1000 are beyond the realm of practical 
loop design. 

Should not exist. 

Fig. 2 C-D Road Design Speed 35 mph 

Minimum Distance D (ft.) 

225 
250 
300 
400 
450 
500 
600 
700 
1000 

Total 20 100.0 

Comments : 

700 ft., subject to weaving criteria (quality of flow II for fig. 1 and 
III rural or IV urban for fig. 2). 



Number Responding 


Percent 


Cumulative % 


1 


5.0 


5.0 


1 


5.0 


10.0 


1 


5.0 


15.0 


3 


15.0 


30.0 


1 


5.0 


35.0 


3 


15.0 


50.0 


5 


25.0 


75.0 


4 


20.0 


95.0 


1 


5.0 


100.0 



B-30 



Percent 


Cumulative % 


5.6 


5.6 


5.6 


11.2 


11.1 


22.3 


16.6 


38.9 


11.1 


50.0 


22.2 


72.2 


11.1 


83.3 


11.1 


94.4 


5.6 


100.0 



Desirable Distance D (ft.) Number Responding Percent 

400 1 

450 1 

500 2 

600 3 

700 2 

800 4 

goo 2 

1000 2 

2000 1 

Total 18 100.0 

Comments : 

The distance is based on weaving volumes (no specific answer given) 

Fig. 2 C-D Road Design Speed 50 mph 

Minimum Distance D (ft.) Number Responding 

400 1 

450 2 

500 3 

600 4 

700 2 

800 5 

900 1 

1000 2 

Total 20 100.0 

Comments : 

500+. If you are referring to long C-D roads that go through several 
local interchanges, then the distances should approach 1400 ft. min. and 2000 
ft. desirable. 

Distances greater than 800-1000 are beyond the realm of practical loop 
design. 

700 ft., subject to weaving criteria (quality of flow II for fig. 1 and 
III rural or IV urban for Fig. 2), 



Percent 


Cumulative % 


5.0 


5.0 


10.0 


15.0 


15.0 


30.0 


20.0 


50.0 


10.0 


60.0 


25.0 


85.0 


5.0 


90.0 


10.0 


100.0 



B-31 



Desirable Distance D (ft.) 

500 
600 
750 
800 

900 
1000 
1200 
1300 
2000 



No. Responding 


Percent 


Cumulative % 


1 


5.6 


5.6 


2 


11.1 


16.7 


1 


5.6 


22.3 


1 


5.6 


27.9 


2 


11.1 


39.0 


5 


27.7 


66.7 


4 


22.1 


88.8 


1 


5.6 


94.4 


1 


5.6 


100.0 



18 



100.0 



Total 

Comments : 

The distance is based on weaving volumes (no specific answer given). 

500+. If you are referring to long C-D roads that go through several 
local interchanges, then the distances should approach 1400 ft. min. and 
2000 ft. desirable. 

Distances greater than 800-1000 are beyond the realm of practical loop 
design. 

General Comments: 



Weaving and volumes will determine length. 

An arbitrary minimum can be chosen but the desirable should be equal 
to or greater than the length required for weaving. 



14. Regarding the values used in answering #13, please indicate below 
whether they differ from values recommended in your state manual. 



Numb er 
Responding 

Percent 



Yes 

5 
27. { 



No 



16.6 



Not Covered 
in Manual 

10 

55.6 



Total Number 
Responding 

18 

100 



Comments : 

No State Manual (N.J.). 

Yes - Manual does not cover fig. 1, but covers fig. 2. 

Yes and not covered in manual (dual response) . 

No or not covered in manual (dual response) . 

Yes, slightly higher than the AASHO Blue Book. 

B-32 



15. What, in your opinion, would be the maximum desirable weaving volume 
in vehicles per hour for the arrangements shown in: 



in Figure 1 



_vph, 



in Figure 2 



vph. 





Figure 1 - Maximum 
Desirable Weaving 
Volume (vph) 



400 

500 

600 

800 
1000 
1100 
1400 
1500 
2000 



Number 
Responding 

2 
2 
2 

1 
2 
5 
1 

1 
1 
1 



Percent 



11 


.1 


11 


1 


11 


1 


5 


6 


11 


1 


27 


6 


5. 


6 


5. 


6 


5. 


6 


5 


6 



Cumulative 


Percent 


11 


.1 


22 


.2 


33 


3 


38 


9 


50 





77 


6 


83. 


2 


88. 


8 


94 


4 


190. 






Total 



18 



100.0 



B-33 



Figure 2 - Maximum 



Desirable 


Weaving 


Number 




Cumulative 


Volume (vph) 


Responding 


Percent 


Percent 


600 




1 


5.9 


5.9 


800 




1 


5.9 


11.8 


, 1000 




4 


23.5 


35.3 


1500 




6 


35.2 


70.5 


1600 




3 


17.7 


88.2 


2000 




2 


11.8 


100.0 


Total 




17 


100.0 




Comments: 











These figures (1000, 2000) could vary depending on whether a heavier 
volume occurred entering or exiting. 

Fig. 1 is not desirable at any volume on a freeway. 

No answer. The maximum weaving volume would depend on the resultant, 
length. As a rule of thumb, I like 1 vehicle per 1 foot of length for 
non-C-D and 1.5 vehicles per foot of length for C-D roads. 

The maximum volume of a loop is 800 vph. This is the primary reason 
loop ramps are not used in Texas. Our new design manual states that all 
loop ramps will have C-D roads. I do not believe that any loop pair can 
actually handle 1600 vph. with or without C-D roads. 

800 and 2000 assuming level grades and a large urban area. 

No answer - depends on length and layout. 

500 and 800 given as answers. D should be known to estimate weaving 
volume. From to 1000 Arph. when D = and D = 1000 respectively. Also 
speed should be known. 

800 and 1000 given as answers. Influence of main roadway weaves or 
"late weaves" would have a volume reduction effect. 

Weaving on freeway mainlines should be avoided (fig. 1=0 vph). On 
C-D roads, maximum weaving volumes will depend on the length of weaving 
section and must be analyzed on a case-by-case basis. 

Volumes will depend on value of D and average speed (500 and 1000 vph 
for 50 mph operating speed) . 



B-34 



16. Regarding the values used In answering question #15, please indicate below 
whether they differ from values recommended in your state manual. 











Not Covered 


Total Number 




Yes 


No 




in Manual 


Responding 


Number 
Responding 


2 


1 




13 


16 


Percent 


12.5 


6. 


2 


81.3 


100 



Comments : 

No State Manual (N.J.). 

Yes - Figure 1 not covered in manual. Figure 2 agrees with manual. 

No and C-D road not covered in manual (dual response) . 



B-35 



17. In designing major interchanges, is uniformity sufficient justifica- 
tion for using the arrangement in Figure 2 at all locations with 
adjacent loop ramps? 









Number 


Responding 


Percent 


Yes 






10 


47.6 


No 






11 


52.4 


Total Number 
Responding 






21 


100. 


Comments for "yes" 


responses : 









Figure 1 is unacceptable. 

The maximum volume of a loop is 800 vph. This is the primary reason 
why loops are not used in Texas. Our new design manual states that all 
loop ramps will have C-D roads. I do not believe that any loop pair can 
actually handle 1600 vph. with or without C-D roads. 

When thinking of standard design, what comes to my mind primarily is 
single exit, single entrance design connected to collector-distributor roads. 
This is a desirable thing to do but is not always feasible in heavily built- 
up urban areas because of the added right-of-way required. 

If costs are not excessive. 

If loop type ramps must be employed. Figure 2 should be minimum acceptable 
design. 

It is desirable to use single exit designs for all interchanges. This 
can be accomplished through the use of C-D roads at cloverleafs. 

Only if uniformity Is being sought to conform with "driver expectancy," 
not uniformity for the sake of uniformity . 



B-36 



Comments for "no" responses: 

I do not think uniformity is an exclusive criteria upon which to make 
this decision. Admittedly, uniformity offers some very significant benefits. 

This condition would usually occur at low volume ramps at a major 
interchange. 

Uniformity of both geometries and signing are important considerations. 
However, each interchange must be designed on its own merit, and at times 
Figure 1 may be appropriate. 

Our studies have led us (Illinois) to conclude that it is not practical 
to do so just for uniformity. There are other priorities which in our (my) 
opinion are more cost effective. 

Where volumes are extremely low and the loop (free flow) design is 
provided only because of a freeway-to-freeway condition, the C-D road should 
be omitted. The C-D roads and related overhead signs can cause confusion 
to the mainline driver, not to mention the cost where not warranted for 
weaving. 

Economics are still involved. 

Loops with weaving should not be used in a major interchange. 

Uniformity is desirable, but with lesser traffic the collector road 
is not required. 



B-37 



18. In your opinion what Is the most appropriate mlnlmuin design speed for 
turning roadways (ramps) on major Interchanges? Please provide an 
answer for both major and minor traffic movement. 



Major movement 
Minor movement 



mph 



mph 



Minimum Des 


Lgn 


Speed 








(mph) 






Number 




Cumulative 


Major Movement 




Responding 


Percent 


Percent 


35 






1 


5^3 


5i3 


40 






1 


5.3 


10.6 


45 






3 


15.8 


26.4 


50 






12 


63.1 


89.5 


60 






2 


10.5 


100.0 



Total 



19 



100.0 



Minimum Design 


Speed 








(mph) 




Number 




Cumulative 


Minor Movement 




Responding 


Percent 


Percent 


25 




2 


10.5 


10.5 


30 




7 


36.9 


47.4 


35 




4 


21.1 


68.5 


40 




2 


10.5 


79.0 


45 




2 


10.5 


89.5 


50 




2 


10.5 


100.0 



Total 



19 



100.0 



Comments : 

50,30 mph — At major interchange higher speeds may be desirable in 
both categories assuming other factors such as right-of-way cost, adverse 
travel distance, additional construction cost, etc. do not out-weigh the 
benefits. 

50,50 mph — for 70 mph on the mainline, the desirable minimum design 
speed for turning roadways should be 60 mph. Obviously, the use of loop 
ramps is ruled out . 

35,20 mph — If directional ramps are used, mlnimums should be 40 mph. 



B-38 



No answer — Urban multi-lane directional ramp ■ 60 mph. 
Urban multi-lane semi-directional ramp = 40 mph 
Rural multi-lane directional ramp = 70 mph 
Rural multi-lane semi-directional ramp = 70 mph 

In rural areas, I believe it is highly desirable to design all multi- 
lane directional ramps to eliminate the need for speed-zoning. 

50,25 mph — Major - 850 ft. radius or 50 mph is considered a good fit 

considering safety, capacity, cost, etc. Minor - 150 ft. radius or 25 mph 

for loops (150 ft. radius would not be advocated on a connection other than 
a iaop) . 

50 and 40 mph would be for directional-type ramps, not loops. Loops 
should be maintained at 225 ft. to 300 ft. radii in the interest of economy 
and uniformity. Providing a 45-50 mph design speed would not be economical 
and would invite drivers to expect the same operating conditions at the 
many 150-225 ft. radii loops in existence. 

No answer - The minimum design speed should be at least 0.7 of the 
main lane design speed. We generally try to obtain higher design speeds than 
minimum. The minimum length of an exit ramp should be based on the minimum 
stopping distance of the main lane design speed. 

50,35 mph - The major movement should be as near the through movement 
as practical. 

50,35 mph - It would be desirable to have the same design speed as the 
mainline on major traffic movements. 

45 to 60 mph for both major and minor movements (45 used in tabulations 
above). For freeway-to-freeway movements, a basic philosophy should be to 
maintain speeds as near to mainline speeds as practical. A 10 mph reduction 
in speed, in my view, could be acceptable. This matter should be related to 
operating speeds in an urban or rural situation. For example, some urban 
areas restrict mainline speeds to 55 mph or less. These comments exclude 
consideration of loop-type ramps. 

45,30 mph - These are minimums based on a freeway design speed of 
70 mph; higher speeds are desirable. 

50,40 mph - The design speed for turning movements should be related 
to the design speed of the through movement. A maximum of 20 mph difference 
in speed should be provided. 

50,50 mph for 70 mph mainline. Use 40 mph for 60 mph mainline and 
30 mph for 50 mph mainline. 

45,30 mph - These speeds are or should be dependent upon the design 
speeds of the through lanes: 

1) Speed of the major movement be no less than 10 mph lower than 

the fastest through lane, but not less than stated above (45 mph). 

2) Speed of the minor movement should be no less than 25 mph lower 

than the fastest through lane, but not less than stated above (30 mph) 



B-39 



19. In your opinion, under which of the following conditions should a left turning 
movement be permitted from the left or high speed lane in a major interchange? 



a. Left turn volume, 
10% of total volume 

b. Left turn volume, 
30% of total volume 

c. Left turn volume, 
50% of total volume 

d. Left turn volume, 
requires 2 lanes 

e. Through numbered 
route turns left 

f. Only alternative 
to a loop ramp 

g. Left turn from right 
lane cost $100,000 
more 

h. Left turn from right 
lane cost $250,000 
more 

i. Left turn from right 
lane cost $500,000 
more 

j . Other (specify) 





Almost 
always 


Sometimes 


Almost 
never 


Total number 
responding 


No. responding 
% 


1 
4.8 


1 

4.8 


19 
90.4 


21 
100 


No. responding 
% 


1 
4.7 


6 

28.6 


14 
66.7 


21 
100 


II 


6 

28.6 


10 

47.6 


5 
23.8 


21 
100 


It 


5 

23.8 


12 

57.1 


4 
19.1 


21 
100 


M 


8 

38,1 


9 

42.9 


4 

19.0 


21 
100 


II 


4 
21.0 


8 

42.1 


7 
36.9 


19 
100 


II 


1 
5.8 


3 

17.7 


13 
76.5 


17 
100 


II 


2 
11.8 


5 
29.4 


10 
58.8 


17 
100 


II 


5 

29.4 


5 

29.4 


7 

41.2 


17 
100 


11 


3 
100 










3 
100 



Other : 

Left turn from right lane in excess of $1,000,000 

Only when a major fork design is used. 

c. above, as the mainline road (thru route continues ahead) 



B-40 



Comments: 

Almost always for c. above assuming that 50 percent would require a 
major fork design, therefore permitted. Almost always for e. above if two 
lanes (or more) are provided. 

If this is a major fork, one must go to the left. But, if main route 
goes left, make this roadway generally straight thru at the fork with the 
other one breaking to the left. 

Sometimes for d. above if the right turn is also two lanes. 

New York's policy statement on left-hand exits and entrances is to not 
do it except for major forks, and then place the higher commercial volume 
on the right. 

g through j are ^zariable based on conditions. 

Any major fork has one on the left. 

Sometimes for c. above assuming the thru route continues ahead. Almost 
never for e. above assuming a ramp connection. 

For g to j — The consideration is mostly cost independent. 

Cost should be an insignificant factor when safety is compromised; that is, 
a left turn ramp should be the only possible alternative. 



B-41 



20. Figures 3 thru 8 in the "Figures Package" alternate exit ramp 
arrangements on one approach of a major interchange. Place an 
X under the use category which characterizes the frequency 
with which your organization uses each arrangement for new, 
i.e., recent and current, designs. 





X 




Z LANts -J 



i l»nts 



O^ 





B-42 













USE CATEGORY 






Figure No. 






Almost Always 
(96-100%) 


Usually 
(66-95%) 


Often 
(36-65%) 


Sometimes 
(6-35%) 


Almost Never 
(0-5%) 


Total Number 
Responding 


3 


No. 


Responding 











3 


14 


17 






% 











17.6 


82.4 


100 


4 


No. 


Responding 











2 


15 


17 






% 











11.8 


88.2 


100 


5 


No. 


Responding 








2 


10 


5 


17 






% 








11.8 


58.8 


29.4 


100 


6 


No. 


Responding 





1 


1 


11 


4 


17 






% 





5.9 


5.9 


64.7 


23.5 


100 


7 


No. 


Responding 





1 


4 


12 





17 






% 





5.9 


23.5 


70.6 





100 


8 


No. 


Responding 


1 


5 


9 


2 





17 






% 


5.9 


29.3 


53.0 


11.8 





100 



Comments : 

Left-hand exits are generally avoided. 

Try to avoid Figures 3 and 4. 

Preference to Figure 8 is initially explored in interchange layouts, 



B-43 



21. The single exit with a fork shown in Figure 8 has been advocated over 
configurations having two separate exits. Presented below are several 
statements regarding the single exit configuration. Please indicate 
the extent to which you agree or disagree with each statement by 
placing a circle around the appropriate word. Following your responses 
to the statements, please list or describe other advantages or dis- 
advantages of the single exit. 




Statement 1. 

The single exit is 
always desirable 
for traffic flow 
advantages . 



No. Responding 
% 







0) 


>. <u 


n 


5>^ 




0) 


iH 0) 


Oi 


H 




M 


00 u 


X> M 


00 


0) 


00 


C oc 


E C 


ti 0) 


0) 


CO 


o td 


3 -H 


O (U 


u 


m 


U CO 


g: -o 


u u 


00 


•H 


4J -H 


C3 


U 00 


< 


O 


CO o 


rH O 


CO <; 








<a Pi 

4J CO 


• 


m 


• 


• 


o <u 


(d 


.Q 


o 


•o 


H P!j 


5 


9 


7 





21 


23.8 


42.9 


33.3 





100 



Statement 2. 

The single exit is 
always desirable 
for safety 



No. Responding 
% 



6 


9 


6 





21 


28.6 


42.8 


28.6 





100 



Statement 3. 

The single exit can 
be used if the exit 
is one lane, but not 
if it is two lanes 



No. Responding 1 13 6 20 
% 5.0 65.0 30.0 100 



B-44 



Other Advantages ; 

If volumes warrant only a one-lane takeoff, signing is simplified. 
If there is a two-lane takeoff required, then a major fork or bifurcation 
can be used. 

My disagreement with statements 1 and 2 above is because of the word 
"always." If the single exit can be designed to accommodate traffic 
volumes, it is desirable. 

When normal routing is involved, it can simplify overall signing for 
the freeway. If a uniform exit pattern is desired, it is compatible. 

Generally prefer #8 due to signing advantages, etc. #5,' 6, 7 are 
acceptable designs and may have advantages in the specific case. 

The single exit greatly simplifies signing. 

Better signing — more time for driver decision. 

As a general comment I do not believe that single lanes do or do 
not help traffic flow. Here again, it would depend on the traffic volumes 
to determine if a single exit or two separate exits should be used. We 
(Texas) have employed both types of design and are pleased with both. 
If a two-lane exit is employed, then one lane must be dropped. We have 
found that signing two separate ramps is less confusing to the driver 
than one exit with a split. 

The design reduces driver confusion. 

Driver expectancy and signing. Merge-diverge turbulence is off the 
main lanes. 

Easier to properly sign the single exit. 

Easier for user, only one choice at the exit. Second decision would 
be at a slower speed. 

Simplified signing design, driver makes one decisive maneuver from 
the mainline, driver hesitation at ramp terminals is lessened. 

Single exits reduce decisions to be made on the mainline to one question; 
to exit or not to exit. Other decisions are made on the lower volume, lower 
speed ramp or C-D road. 



B-45 



Other Disadvantages: 

Signing of the single exit is critical. 

The volume of traffic using the ramp and the maximum distance that can be 
obtained between the two ramp exits will bear on whether a single exit will 
function better than two exits. 

Where complex routing is involved, it can overload signs. When a two- 
land exit is used, it creates a weaving section. 

The problems to be weighed in the balance is that a multilane or 
major fork design is often necessary. This is more complicated than 
two single exits. 

Capacity restraint. 

Single exit increases construction and right-of-way costs. 

Second fork problems cause backups, large spatial requirements. 

Design, right-of-way, and topography may not be compatible with 
resulting high increase in cost. 

With high volumes, the single exit may be a problem due to concentra- 
tion of two ramp flows at one exit. 

Cost, spatial requirements. 

Single exit designs are usually more expensive due to the need for 
additional and/or wider structures. 



B-46 



22. If the turning volume In Figure 8 requires a two-lane exit, would one 
of the other arrangements be more desirable? If yes, circle the more 
desirable arrangements as depicted in Figures 3-7. If you circle more 
than one of the alternative arrangements, indicate the most desirable 
by putting an "a" below it, the next most desirable a "b", etc. 



a. Yes 



b. No 



Figures 




Nunfcer Responding 
Percent 



Figure 
Desirability 



Yes 

8 
38.1 

4 

b c 



No 
13 
61.9 



Total No. 
Responding 

21 

100.0 



No . Respond- 
Ine 



10 
Comments for "no" responses: 



400 121 221 



We have found out that signing two separate ramps is less confusing 
to the driver than one exit with a split. The answer depends a great 
deal on whether you have adequate D distance to provide the Figure 5 
arrangement. If you do not, then use the arrangement in Figure 8. 

Figures 6 and 7 are sometimes more desirable. 



B-47 



23. For each of the exit arrangements shovm in Figures 3-8, Indicate the 
minimum and desirable distance D between exit noses. Your answer Is 
to reflect your opinion and not necessarily the current practice of 
your organization. For the purpose of answering you should assume a 
flat grade and the existence of adequate signing. 




2 lANti- 




X 




i L*Nca 



I t*Mrs -J 





FIGURE 3 



Minimum Distance D (ft.) Number Responding 



350 

500 

600 

800 

900 

1000 

1200 

1500 



Percent 


Cumulative % 


6.7 


6.7 


13.3 


20.0 


33.3 


53.3 


13.3 


66.6 


6.7 


73.3 


6.7 


80.0 


6.7 


86.7 


13.3 


100.0 



Total 



15 



100.0 



B-48 



Desirable Distance D (ft.) Number Responding 



600 
1000 
1200 
1400 
2000 



Percent 


Cumulative % 


12.5 


12.5 


37.5 


50.0 


18.8 


68.8 


6.2 


75.0 


25.0 


100.0 



Total 



16 



100.0 



FIGURE 4 



Minimum Distance D (ft.) Number Responding 



350 

500 

600 

800 

1000 

1200 

1500 



Percent 


Cumulative % 


6.7 


6.7 


13.3 


20.0 


40.0 


60.0 


13.3 


73.3 


6.7 


80.0 


13.3 


93.3 


6.7 


100.0 



Total 



15 



100.0 



Desirable Distance D (ft.) Number Responding 



600 
900 
1000 
1200 
1400 
2000 

Total 



2 
1 
6 
2 

1 
4 

16 



FIGURE 5 



Minimum Distance D (ft.) Number Responding 



350 
400 
500 
600 
670 
800 
1000 
1400 

Total 



1 
1 
1 

4 
1 
5 
7 
1 

21 



Percent 


Cumulative % 


12,5 


12.5 


6.3 


18.8 


374. 


56.2 


12.5 


68.7 


6.3 


75.0 


25.0 


100.00 



100.0 



Percent 


Cumulative % 


4.8 


4.8 


4.8 


9.6 


4.8 


14.4 


19.0 


33.4 


4.8 


38.2 


23.7 


61.9 


33.3 


95.2 


4.8 


100.0 



100.0 



B-49 



Desirable Distance D (ft.) Number Responding 



600 
700 
900 
1000 
1200 
1500 
1600 
2000 
2300 



Percent 


Cumulative % 


4.8 


4.8 


4.8 


9.6 


14.3 


23.9 


23.8 


47.7 


9.5 


57.2 


19.0 


76.2 


9.5 


85.7 


9.5 


95.2 


4,8 


100.0 



Total 



21 



100.0 



FIGURE 6 



Minimum Distance D (ft.) 

350 

400 

500 

600 

670 

800 
1000 
1400 



Number Responding 

1 
1 
1 
4 
1 
5 
7 
1 



Percent 


Cumulative % 


4.8 


4.8 


4.8 


9.6 


4.8 


14.4 


19.0 


33.4 


4.8 


38.2 


23.7 


61.9 


33.3 


95.2 


4.8 


100.00 



Total 



21 



100.0 



Desirable Distance D (ft.) 

600 

700 

900 
1000 
1200 
1500 
1600 
2000 
2300 



Percent 


Cumulative % 


4.8 


4.8 


4.8 


9.6 


14.3 


23.9 


23.8 


47.7 


9.5 


57.2 


19.0 


76.2 


9.5 


85.7 


9.5 


95.2 


4.8 


100.0 



Total 



21 



100.0 



B-50 



.i 



FIGURE 7 



Minimum Distance D (ft.) 

400 

500 

600 

670 

800 

900 
1000 
1200 
1500 
2300 



Number Responding 

1 
1 
3 
1 
5 
2 
4 
2 
1 
1 



Percent 


Cumulative % 


4.8 


4.8 


4.8 


9.6 


14.3 


23.9 


4.8 


28.7 


23.7 


52.4 


9.5 


61.9 


19.0 


80.9 


9.5 


90.4 


4.8 


95.2 


4.8 


100.0 



Total 



21 



100.0 



Desirable Distance D (ft.) 

700 

900 
1000 
1200 
1500 
1600 
2000 
2500 



Number Responding 


Percent 


Cumulative % 


1 




5.0 


5.0 


3 




15.0 


20.0 


5 




25.0 


45.0 


2 




10.0 


55.0 


4 




20.0 


75.0 


2 




10.0 


85.0 


2 




10.0 


95.0 


1 




5.0 


100.0 



Total 



20 



100.0 



Minimum Distance D (ft.) 

400 
500 
600 
800 
900 
1000 



FIGURE 


8 






Number Responding 


Percent 


Cumulative % 


1 




4.9 


4.9 


2 




9.5 


14.4 


7 




33.3 


47.7 


7 




33.3 


81.0 


2 




9.5 


90.5 


2 




9.5 


100.0 



Total 



21 



100.0 



B-51 



Percent 


Cumulative 


J % 


4.8 


4.8 




14.3 


19.1 




9.5 


28.6 




23.7 


52.3 




33.3 


85.6 




4.8 


90.4 




4.8 


95.2 




4.8 


100.0 





Desirable Distance D (ft.) Number Responding 

700 1 

800 3 

900 2 

1000 5 

1200 7 

1400 1 

1500 1 

1600 1 

Total 21 100.0 



Comments : 

The distances are general. Physical conditions and economics could play 
an important part in determining length. They would not vary too much from 
above in most instances. 

Figures 3 and 4 are not recommended. 

It is assumed that these values are desirable minimums since generally 
the greater the distance the better the overall situation (for all desirable 
distances) . 

Existence of adequate signing should be the main factor in determining 
the minimum distance. No answer for Figures 3 and 4; left-hand ramps should 
not be built on future projects. 

No answer for Figures 3 and 4; they should not be considered regardless 
of "D" distance. 

Distances are not given for Figures 3 and 4 because, in my opinion, 
they should never be used. 

Fig. 3-7, min. D = 800-1000 ft, (800 ft. used in tabulations) 

Fig. 3-7, desirable D = 1000-1500 ft. (1000 ft. used in tabulations) 

Adequate signing is a key feature. 

Left-hand ramp exit situations should recognize lane changes and 
indecisiveness of drivers. 

Fig. 5-7, min. D = 670 ft., length of standard exit ramp taper. Since 
I believe that left-hand exit ramps should never be used, I have not shown 
distances for Figures 3 and 4. Distance for Figure 8 (600 and 900) is for signing. 

Distances are given assuming a mainline design speed of 70 mph and based 
on the assumption that no more than one decision be made within 10 sec. 

Fig. 7, desirable D; should not exist. 



B-52 



i 



24. Does your opinion, as expressed in answering question #23, differ 
from the standard practices used in your organization? 

Number Responding Percent 

Yes 6 37.5 

No 10 62.5 

Total 16 100.0 

Comments. for "yes" responses: 

Left-hand ramps should not be built on future projects. We have not 
been 100% successful in avoiding left-hand ramps, but this is our objective. 

We are generally not able to get the distances shown in #23. 



Comments for "no" responses: 

The distances in #23 are general. Physical conditions and economics 
could play an important part in determining length. They would not 
vary too much from those given in #23 in most instances. 

We have recommended only one value and recognize it is a desirable 
minimum distance. 



General Comments: 

Since each situation is evaluated separately, we have no standard 
practice for this situation. 



B-53 



25. Please state your opinion as to the conditions under which entrances 
from the left should be permitted. 

Note: Answers with "Major fork," probably should be branch connections. 

Conditons : 

a. High relative traffic volumes (when nearly equal or greater left 
turning volume) . 

b. High speed merge - auxiliary lane, etc. 

c. Improves downstream weaving situation. 

d. Economical (cost is prohibitive). 

e. No merging areas, all entrance lane(s) is/are added to mainline. 

f. At least one entrance lane is added. 

g. Horizontal and vertical geometry must be conducive to proper merge. 

h. Parallel auxiliary lane is at least 2000 feet. 

i. When the traffic volume warrants the addition of another through 
lane. 

j . If there is no other way to construct a ramp and sufficient 
warrants for the movement (over 250 vph) . 

k. At all major forks. (Related to 'a'.) 

1. No downstream exits on the right near enough to cause weaving. 

m. Where necessary number of entrance lanes cannot be provided 
on right. (Related to 'j ') 

n. Not compounded by right hand entrance in same proximity. 

o. Under no conditions because of safety hazard, e.g., "blind 
spots" in view of the through traffic. 



B-54 



26. 



Shown in Figures 11 and 12 are two arrangements for handling the 
situation where two ramps enter the main roadway from the right. 
Please put an"'x" beside the arrangement you feel is the most 
desirable from the standpoint of safety and operations. 




Two successive entrances to the through roadway 



-Z Lamps 




Merging of two ramps to provide a single entrance 
to the main line. 



Figure 



# Responding 



Percent 



11 
12 



1 
18 



5.2 
94.8 



Total 



19 



Comments : 



Cannot answer 



Answer depends on traffic volumes, weaves, etc. 



Both are good arrangements and Fig. 11 would be OK if aux. was added. 
The addition of aux. lane would of course depend on the length of L. Fig. 12 
would be safer from a "min. point of access" criteria, however, we have used 
both and the criteria that determines which to use was traffic volumes and 
available space. 

B-55 



Either can serve. 

Depending on the entering volume. 



27. Considering only current (or recently completed) design efforts, please 
describe briefly the primary conditions and/or situations which lead to 
the use of the clover leaf configuration for major interchanges. 

Conditions : 

a. Clover leaf is cheapest, therefore, it is the obvious choice in 
rural, low volume situations. We upgrade when needs call for 
it. (Calif.) 

b. When cost too high for direct or semi-direct ramps. ('a') 

c. None, cloverleaf not used in Texas for new major interchanges. 
(Georgia, New Jersey 

d. Moderate turning movements requiring only single lane ramps. 
(Ohio 800 dhv or less) 

e. Right-of-way (No space restrictions) 

f. Topography 

g. Light weaving volumes. 

h. When neither route turns. 

1. Stage construction is easily accomplished with cloverleaf. 

j. Highly directional flow so adjacent loop ramps never carry maximum 
volumes simultaneously. 

k. Use of collector-distributor roads to eliminate weaves. 

1. Right-angle of intersection. 



B-56 



28. Figures 9 thru 12 indicate four alternate entrance ramp arrangements 
on one leg of a major interchange. Place an x under the use category 
which characterizes the frequency with which your organization uses 
each arrangement in current (or recent past) designs. 



K 



1 



■2'i.AHes 



i 



'Z Lases 




11 



|— £ La^/fs 




-2 L^fcs 




Figure Ntimber 



o 
u o 

in iH 

O I 

e vD 

< ^ 



USE CATEGORY 



H 0\ 

CO I 

zi vo 

CO vO 



<u I 
4-1 <o 

o ■^ 



CO 

S --^ 
<u n 

e I 

O NO 
CO '-' 



0) 

o m 
S i 

tH O 



00 

c 

•H 

c 
o 
o. 

CO 
(U 



No . Responding 
% 



17 
100 



17 
100 



10 



No . Responding 
% 



17 
100 



17 
100 



11 



No . Responding 
% 



1 
5.9 



8 
47.0 



6 
35.3 



2 17 
11.8 100 



12 



No . Responding 



1 
5.9 



6 

35.3 



8 

47.0 



2 

11.8 



17 
100 



Comments : 

Try to avoid Figures 9 and 10. 






B-57 



29. If an extra lane is added to the through roadway as a continuation 
of the entrance ramp from the left would your answer to question 
//28 change? 





Yes 


No 


Total Number Responding 


Number 
Responding 


13 


4 


17 


Percent 


76.5 


23.5 


100 



Comments for "yes" responses: 

With a continuous free lane the left-hand entrance becomes more accept- 
able. Multiple right and left entrances in the same vicinity would adversely 
affect operations. Ramp entrances and exits which create compound weaving 
situations would have to be avoided. 

When no merge is involved the left entrance would be acceptable. 

It is still more desirable to have the ramp on the right. 

The continuous extra lane for the left entrance ramp would allow the 
left entrance to be more acceptable and used more often by reducing or 
eliminating traffic in the left lane from merging right in a short distance. 

Usage of Figures 9 and 10 would probably increase since we would not 
be as hesitant to bring a ramp in on the left if no merge is required, and 
the overall cost of the interchange would be notably reduced. 

No objection to a left on-ramp if a lane is added. 

It would eliminate part of the merging problem. 

Continuation of the left entrance ramp as a through lane prevents the 
hazardous merging from left to right into the high-speed lane. 

An extra lane is required if the ramp is on the left. 

Yes; however, entrance ramps from the left should not be a matter of 
design practice. 

Comments for "no" responses: 

No left ramps should be provided. 

Left entrances should never be provided unless the two legs of the 
connection are of equal importance and have an equal number of lanes. 

Left-hand entrances are always hazardous, especially for trucks, 
since the following exit will generally be on the right, and thus a 
merge to the side of reduced visibility must eventually be made. 



B-58 



30. If the turning volume in Figure 12 requires a two-lane entrance, would 
one of the other arrangements be more desirable? (Circle one) If 
yes, circle the more desirable arrangements as depicted in Figures 9, 
10, and 11. 



a. Yes 



b. no 



Figures 9 10 



11 




U 



r-^Lfif/n 




: — £" Zitwra 



10 








Yes 




No 


Total Number 
Responding 


Number Respon 


ding 


9 




12 


21 


Percent 




42.9 
Figure 
9 10 


No. 
11 


57.1 


100 


Number 

Responding 







9 






Percent 







100 







Comments for "yes" responses: 

Each situation requires individual analysis . 

The answer is "no" if an added lane or auxiliary lane is available. 
The answer is "yes" if no additional lane is provided. 

Figure 11 if adequate distance for L is available. If not, use Figure 12 
and add two lanes. 



B-59 



Figure 11, if L were at least 2,500 ft. 

Comments for "no" responses: 

Either 11 or 12 can be designed suitably. 
An extra lane would be picked up. 



I 



B-60 



31. For each of the four entrance arrangements shown in Figure 9-12 
indicate your opinion as to the minimuTH and desirable distance L 
between entrance noses. 




1 '-I? Lanes 



/' 




f'^Lfi-JF.b 




11 



JT- 



^ LaN£S 




-2 LahE3 



10 




Fig. 9 Minimum Length (L) 



L (in feet) 

500 

575 

900 
1000 
1200 
1400 
1500 

Total 



No. Responding 

1 
1 
2 
8 
2 
1 
3 

18 



Fig. 9 Desirable Length (L) 



Percent 


Cumulative % 


5.6 


5.6 


5.6 


11.2 


11.1 


22.3 


44.5 


66.8 


11.1 


77.9 


5.6 


83.5 


16.7 


100.0 



L (in feet) 


No . Re spoil 


ding 


Percent 


Cumulative % 


900 


1 




5.9 


5.9 


1200 


2 




11.8 


17.7 


1400 


1 




5.9 


23.5 


1500 


2 




11.8 


35.4 


1600 


1 




5.9 


41.3 


1800 


1 




5.9 


47.2 


2000 


8 




47.1 


94.3 


2400 


1 




5.9 


100.0 



Total 



17 



B-61 



Fig. 10 Minimum Length (L) 



L (In feet) 


No. 


Responding 


Percent 


Cumulative % 


500 




1 


5.6 


5.6 


575 




1 


5.6 


11.2 


900 




2 


11.1 


22.3 


1000 




9 


50.0 


72.3 


1200 




2 


11.1 


83.4 


1500 




3 


16.7 


100.0 



Total 



18 



L (in feet) 

900 

1200 
1400 
1500 
1600 
1800 
2000 

Total 



Fig. 10 Desirable Length (L) 
No« Responding 



Percent 


Cumulative % 


5.9 


5.9 


11.8 


17.7 


5.9 


23.6 


11.8 


35.4 


5.9 


41.3 


5.9 


97.2 


53.0 


100.0 



17 



L (in feet) 

500 

575 

900 
1000 
1150 
1200 
1400 
1500 
1600 
2300 

Total 



Fig. 11 Minimum Length (L) 
No. Responding 



Percent 


Cumulative % 


4.8 


4.8 


4.8 


9.6 


14.3 


23.9 


23.8 


47.7 


4.8 


52.5 


14.3 


66.8 


14.3 


81.1 


9.5 


90.6 


4.8 


95.4 


4.8 


100.0 



I 



21 



£-62 



Fig. 11 Desirable Length (L) 



L (In feet) 

900 
1000 
1150 
1200 
1500 
1800 
2000 
2A00 
2500 

Total 



L (In feet) 

400 

500 

600 

750 

800 

900 
1000 
1200 
1500 
1600 

Total 



L (In feet) 

600 

700 

800 

900 
1000 
1200 
lAOO 
1500 
2000 

Total 



No . Responding 

1 
2 
1 

2 
3 

1 
7 
1 
1 

19 



Fig. 12 Minimum Length (L) 
No. Responding 



3 
3 
3 
1 
5 
2 
1 
1 
1 
1 

21 



Fig. 12 Desirable Length (L) 



No. Responding 

1 
1 
2 
3 
1 
7 
1 
1 
3 

20 



Percent 


Cumulative % 


5.3 


5.3 


10.51 


15.8 


5.3 


21.1 


10.5 


31.6 


15.8 


47.4 


5.3 


52.7 


36.9 


89.6 


5.3 


94.9 


5.3 


100.0 



Percent 


Cumulative % 


14.3 


14.3 


14.3 


28.6 


14.3 


42.9 


4.8 


47.7 


23.8 


71.5 


9.5 


81.0 


4.8 


85.8 


4.8 


90.6 


4.8 


95.4 


4.8 


100.0 



Percent 


Cumulative % 


5.0 


5.0 


5.0 


10.0 


10.0 


20.0 


15.0 


35.0 


5.0 


40.0 


35.0 


75.0 


5.0 


80.0 


5.0 


85.0 


115 JO 


100.0 



B-63 



Comments : 

Fig. 9-12 (min. length) are +. 

Fig, 9-12 (desirable length) - cannot answer - depends on other factors 
- merging distance //12 depends upon approach speeds, etc. 

Fig. 9 and 10 (Min. & desirable lengths) - not applicable 

Fig. 9 and 10 — undesirable condition 

Fig. 11 (min. & desirable) — length of standard entrance taper 
(New York) . 

1000 ft. min. length for all figures. 

2000 ft. desirable length for all figures 

Fig. 9 and 10 ~ 

Fig. 11, desirable length — "should not exist" 



32. Does your opinion, as expressed in answering question #31, differ 
from the values recommended in your state manual? 



Response 


No . Responding 


Percent 


Yes 


1 




7.2 


No 


2 




14,3 


Not Covered 








in State Manual 


11 




78.6 


Total 


14 







I 



Comments for "no" responses: 

Only Fig. 11 and 12 covered in state manual (Ohio) 



I 



B-64 



33. In designing entrance ramps you may prefer the taper type acceleration 
lane for one situation or condition and the parallel type for another. 
Please state below the situations or conditions under which you prefer 
each type of geometry. 

The tapered type is preferred unless other considerations indicate the 
desirability of a more sophisticated treatment. A parallel type is recom- 
mended where merging distance is believed necessary, especially where anti- 
cipated through volumes show minimum merging opportunity or where other 
adverse conditions indicate the desirability of this treatment. 

Tapered type is preferred. 

Tapered type is always preferred except where weaving is involved. 

Prefer the tapered tjrpe unless ramp is 2 lanes on the right with an 
added lane, or unless the ramp is on the left. 

Tapered type entrances are desirable under most conditions where 

sight distance, grades, and alignment are acceptable. Parallel type 

entrances may often be used where longer acceleration distances are 
desired, and where continuous lanes are to be used. 

Tapered type preferred for all conditions but- especially on horizontal 
tangents . 

We generally provide a standard tapered acceleration lane (L ~ 1000 ft. +) 
Additional parallel lanes are added as traffic requires. 

The tapered type is most always preferred simply for the sake of uni- 
formity and as recommended by AASHO since the Special Freeway Committee Report 
of I960 (Red Booklet) . Parallel type used when situations warrant a climbing 
lane as a part of an entrance, but a 50:1 taper would still be the optimum 
desired at the terminal. 

The tapered type is usually used except on an upgrade or a two-lane 
of f -ramp. 

We employ the taper type of ramp in most cases because our (Texas) 
experience has shown that most drivers drive a parallel acceleration ramp 
as if it was a taper ramp. 

A tapered type is used for most cases. When an upgrade exists beyond 
the entrance, a parallel deceleration lane may be needed for commercial 
vehicles to attain a reasonable operating speed before entering the mainline. 

We do not use a tapered type in present design. A parallel type is used 
all the time. 

Either alternative can be designed to work well. 

Tapered type used in all cases except where grade sight distance and 
capacity would require a length longer than practical for a taper. 



B-65 



Tapered type is favored for all conditions since the merge is gradual. 

The taper-type is adequate if there is plenty of length (1000 ft. +) . 
Parallel may be used if lengths is restricted and/or if there is very heavy 
mainline traffic. 

Tapered type is generally preferred in urban high volume situations 
principally to encourage the driver to be aggressive. Parallel type is 
generally preferred on high speed-medium volume facilities where inter- 
changes are spaced at or greater than two-mile intervals, such as toll 
roads or rural remote interstate routes. It is also preferred for two-lane 
entrance situations and trucks entering on an upgrade. 

Tapered type used in all cases except: 

1. At an entrance carrying a large truck volume in combination 
with a steep upgrade. The lane should then be designed as a 
climbing lane. 

2. For lane balance, when it is desired to maintain the basic 
number of through freeway lanes at a two-lane entrance. 

3. On highways curbed at the edge of the travel lane. 

I prefer the tapered type where the ramp merges at a flat angle and 
sight distance is good. Parallel used when the reverse is true. 

In rural situations where the mainline traffic volumes are low, a 
taper is used. In most situations a parallel type is used because the 
parallel type speed-change lane will allow both the mainline driver and the 
entering driver to adjust to the final entering maneuver at the prevailing 
mainline operating speed. 



B-66 



34. In designing exit ramps you may prefer the taper type deceleration 

lane for one situation or condition and the parallel type for another. 
Please state below the situations or conditions under which you 
prefer each type of geometry. 

The tapered type is normally used. A parallel-type is used under very 
high exiting volumes, or where geometry indicates desirability in order to 
overcome deficiencies in target value or the operational characteristics of 
a normal taper. 

The tapered type is preferred. Where there is the possibility of a 
back up on the mainline a parallel type is used. 

Tapered type used for a major fork or a major terminal where little or 
no slowing is required. The parallel type is used exclusively where decel- 
eration is required. The beginning is abrupt and well-defined, thus alerting 
the motorist that the nose is ahead. 

Tapered type is preferred in most situations. A parallel lane is pre- 
ferred if the off-ramp is 2-lane or if the ramp terminal is such that decel- 
eration or storage back onto the freeway is needed. Also used if stopping 
sight distance is restricted. 

Tapered type exits are preferred where high-speed exits are used. 
Parallel type exits are preferred where slower ramp speeds are necessary. 

Tapered type used on all tangents, parallel type on all curves. 

A standard taper is used for freeway-to-freeway exits. Approach auxiliary 
lanes are provided as traffic requires. 

Tapered type is desired at 3°-5° rr 15:1 angles of exit to fit the path 
that most drivers take and to show a definite point of take-off from the 
mainline. Longer tapers cause problems, for through drivers make uninten- 
tional exits, especially during poor visibility conditions at night. The 
parallel type is adequate for use where the ramp must depart from the main- 
line on a horizontal curve, (This condition"^ should be avoided if at all 
possible.) A parallel lane with an abrupt take-off is good to distinguish 
the ramp from the mainline roadway and avoid the problem described above. 

The tapered type is usually used except on a downgrade or a 2-lane off- 
ramp. 

We (Texas) employ the tapered type in most cases because our experience 
has shown that most drivers drive a parallel ramp as if it was a taper 
ramp. 

The tapered type is used for most cases, A parallel type is used for 
a mainline curve to the left to reduce the appearance of the mainline going 
up the ramp. It is also used where an exit is unavoidably hidden beyond 
a crest vertical curve with the parallel lane starting prior to the crest. 



B-67 



We find that the tapered section is not used. A parallel type with an 
80 ft. taper is employed all the time. 

Either can be designed to work well. 

The taper is favored in all instances except where the difference in 
ramp and freeway speeds are so great that the taper would not be practical, 
and where sight distance is a problem. 

The parallel type is favored for all situations. It can be used as a 
taper or a lane for reducing speed and also provides a storage area if 
required during excessive traffic problems. 

A tapered type is adequate if the ramp permits an exit at 60-55 mph. 
Too many tapers are used where braking on the taper (thus in mainline) is 
necessary due to a short finger or a tight loop ramp ahead. In this situa- 
tion a parallel type is better. 

A tapered type is preferred under medium traffic situations and high 
ramp design speeds. Parallel preferred under dense mainline traffic condi- 
tions — heavy rajip volumes, either urban or rural situation and relatively 
low ramp design speeds. It is also preferred fcr a 2-lane exit situation. 

Tapered type is preferred for all cases except : 

1. When a right-hane exit is unavoidably located on a mainline 
curve to the left, and it is feared that the use of a taper 
might result in inadvertent use of the ramp by through traffic. 
Note that this configuration is undesirable and should be 
avoided. 

2. When an exit is unavoidably located immediately beyond a crest 
vertical curve or in any other area of restricted visibility. 

3. For lane balance, when it is desired to maintain the basic number 
of through freeway lanes at a 2-lane exit. 

I prefer the parallel type in all cases, if it is long enough and signed 
adequately to encourage drivers to use it for deceleration. 

A tapered type is favored in most situations since proper planning should 
allow the designer to build adequate storage space into this terminal facility 
to prevent slowdown on the mainline facility. It gets the exiting driver 
away from the mainline the fastest and does not present a confusing situation 
to the through driver. When it is not feasible to build adequate storage 
space into tapered ramps, a parallel type is favored. It does get the driver 
out of the through stream, and it does present a confusing situation to 
the through driver as it will appear the facility has added a lane. 



B-68 



35. Figures 13 through 22 show alternative methods for lane drops when 

turning volumes justify a reduction in the number of through traffic 
lanes. Place an x under the use category which characterizes the 
frequency with which you organization currently use each arrangement. 
(Note: neglect the shape of the deceleration lane and nose geometry.) 



-3 L Anes 



3 LA/tei-r 



^L> 



2 LA»ts- 



13 



Dpop Laaif 
BrroNO /.oructi/iMci 




^ LAH£i 



Ditop Ltrr l*Hr . 




SL 



3lAnrs 



2 Lamcs- 




^4 Laucs 



JlAMft-^ 




^^ Lancs 



Drop Le^r Lanc 



z. 



3lANCi- 



-4 LANfi 



■1 LANCi-. 



19 



20 



Dkop Lanc 
BcroHO IuniKHAr<St 



■4 LAriti 



DifOP Ltrr lAnt- 



3 (AUCS- 




B-69 







USE CATEGORY 








Figure 
Number 


Almost Always 
(96-100%) 


Usually 
(66-95%) 


Often 
(36-65%) 


Sometimes 
(6-35%) 


Almost Never 
(0-5%) 


CO 

o 

H 


13 Number Responding 
% 


2 
11.8 


5 
29.4 


1 
5.9 


4 
23.5 


5 
29.4 


17 
100 


14 Number Responding 
% 


1 
5.9 


4 
23.5 


2 
11.8 


8 
47.1 


2 
11.8 


17 
100 


15 Number Responding 
% 














3 

17.7 


14 . 
82.4 


17 
100 


16 Number Responding 
% 


1 

5.9 


3 

17.7 


3 

17.7 


9 
53.0 


1 
5.9 


17 
100 


17 Number Responding 
% 


1 
5.9 


4 
23.5 


3 

17.7 


4 
23.5 


5 

29.4 


17 
100 


18 Number Responding 
% 


1 
5.9 


4 
23.5 


3 

17.7 


7 
41.2 


2 
11.8 


17 
100 


19 Number Responding 
% 














3 

17.7 


14 
82.4 


17 
100 


20 Number Responding 
% 






2 
11.8 


3 

17.7 


3 

17.7 


9 

53.0 


17 
100 


21 Number Responding 
% 


2 
11.8 


7 
41.2 


4 
23.5 


2 
11.8 


2 

11.8 


17 
100 


22 Number Responding 
% 














2 

11.8 


15 

88.2 


17 
100 


Comments: 















The case that New York State advocated is not indicated. That is, 
drop the left lane well beyond the interchange where sight distance is 
unquestionably good. We would use an arrangement similar to Fig. 19 
with the lane drop 1/2 mile beyond the last ramp. 

Fig. 20 - won't work as a two-lane exit. 

Fig. 13 & 17 - very desirable — although hard to obtain if inter- 
changes are closely spaced. Fig. 16 - very desirable. 

Fig. 15 & 18 - Personally, I am not opposed to this method if full 
width lane is not carried beyond nose. 



B-70 



36. In your opinion, when a lane is dropped beyond the interchange, which 
lane should be dropped? (Circle one) 

Namber Responding 

Left Lane 3 

Right Lane 16 

Either _J^ 

20 100 

Comments : 

Left Lane: 



Percent 


15 





80 





5 


.0 



Begin the lane drop taper 1/2 mile beyond the last ramp of an inter- 
change and use "design speed" x "lane width" for taper length. 

Prefer not to have the slower moving vehicles generally in the right 
lane to move over into the faster traffic. 

Reasons for the choice: 

1) Lane can be extended at a later time without disrupting inter- 
change ramp terminals and bridge pier spacing. 

2) Median transition (widening) occurs almost automatically without 
need for dog legs in alignment (Medians usually are widened at 
same location of lane drop) . 

3) Since drivers usually stay to the right, fewer have to change 
lanes for a left side lane drop. 

Either right or left side designs can work OK if designed with good sight 
distance, proper recovery area beyond the taper, and far enough removed from 
other points of conflict such as ramp terminals. 

Either Lane: 



Depends on geometries at lane drop and traffic distribution. 

Right Lane: 

Safer to merge right lane 

On two-lane (one-way), we have wider paved shoulder on right. Also, 
I would rather deal with volume than speed. On three- or four-lane (one- 
way) , the shoulders are the same width and the volume is more evenly dis- 
tributed but speed is higher on left, therefore, prefer right-hand drop. 

Keeps speed changes and weaving on the right side where drivers 
expect it. 

This is in conflict with official department policy which states that 
left lane should be dropped since it carries lower volume. I maintain 
that it is preferable to drop the lane carrying lower speed traffic. 

Left lane usually higher speed and driver has generally poorer visi- 
bility of the merging operation. 



B-71 



The left lane is commonly used more for high speed through traffic 
passing the Interchange and should not have a lane drop. The right lane 
is used by the interchanging traffic which is more desirable for lane 
drop. 

High speed traffic is in the left lane and should not be disrupted 
by a lane drop. 

Through traffic in high-speed left lanes should not be operationally 
Interrupted. 

Driver expectancy. 

Better operation and greater safety for the higher speed 'traffic in 
the remaining lanes. 

The burden of caution should be on the slower driver who generally 
occupies the right lane; and also, he has the least possibility of vehicle 
"blind spots" as he is closest to the through traffic. 



I 



B-72 



37. Where the turning traffic volume requires a two-lane turning roadway, 
certain geometries may be less desirable than others if operational 
problems are to be minimized. Indicate the frequency with which 
your organization currently uses each of the features listed below 
on 2-lane turning roadways in a major interchange. 



CO 

to 

<! 
+j 

CO 

o 

r-t 

< 



USE CATEGORY 



I 

CO vD 



m 

(U I 

M-l CO 

o ^^ 



CO 

(U 

6 ^ 

•H &-? 
0) CO 



6 
o 

C/3 



(U 

<u 

CO 5^ 

o to 



Comments : 

50 mph for part e (often) . 



iH O 

CO (O, 

•U CO 

O d) 

H Pi 



a) 


Two-lane exit roadway 


No. 

Responding 
% 


2 
11. 


8 


3 
17.7 


4 
23.5 


7 
41.1 


1 
5.9 


17 
100 


b) 


One-lane exit followed 
by widening to 2-lanes 


No. 

Responding 
% 








2 
11.8 


2 
11.8 


6 
35.3 


7 
41.7 


17 
100 


c) 


Two-lane entrance 
roadway 


No. 

Responding 
% 


1 
5. 


9 


2 
11.8 


4 
23.5 


6 
35.3 


4 
23.5 


17 
100 


d) 


Two lanes merged to 
one lane before 
entrance 


No. 

Responding 
% 


1 
5. 


9 


3 
17.7 


4 
23.5 


5 
29.4 


4 
23.5 


17 
100 


e) 


Design speed of turn- 
ing roadway not less 
than 70% of mainline 


No. 

Responding 
% 


2 
11. 


8 


8 
47.0 


3 

17.7 


4 
23.5 







17 
100 


f) 


Design speed of turn- 
ing roadway between 
50% and 69% of 
mainline 


No. 

Responding 
% 








3 

17.7 


2 
11.8 


10 
58.7 


2 
11.8 


17 
100 



Almost never use two-lane entrance roadway (c) . 
this will operate as one wide lane. 



Unless two lanes are added 



B-73 



38. Describe the more Important geometries which you feel should be 

avoided on two-lane ramps if operational problems are to be minimized. 

Geometries to be avoided: 

a. Stopping sight distance restrictions to decision points. 

b . Two lanes at entrances and exits unless mainline auxiliary lanes are 
provided. 

c. Improper combination of horizontal and vertical curves, restrictive 
alignment . 

d. Mainline lane drop unless high percentage of traffic turns, (Exit?). 

e. Abrupt tapers and sharp radii, design speeds of less than 50 mph. 

f . Less than two lanes at entrance ramp gore or it will operate as 
one wide lane. 

g. Two lanes at exit ramp gore, drop one lane at the gore, 
h. Steep grades. 

i. Short auxiliary lanes, should be 2000-2500 ft. for capacity. 

j. Hidden exits and entrances (related to a, e, h) . 

k. Jointing and striping which do not coincide. 

1. Jointing which requires crossing by preference or major roadway 
traffic. 

m. Two lanes merged to one lane on curve before entrance (related to f) . 

n. Two lane ramps, if possible. 

o. Lack of distance between decision points. 

p. Inadequate signing and related traffic control. 

q. No escape lane ahead of nose along freeway. 

r. Curb at edges of pavement. 

s. Proximity of adequate structures. 

t. Successive entrance and exit conditions. 

u. Lack of lane balance. 



B-74 



V. Use of two lanes on loop ramps. 

w. Geometries of lesser order than single lane ramp. 

X. Providing two lanes at exit and entrance ramp terminals. (Two 
lanes only on ramp , branch out to two for exit , back down for 
entrance) . 



39. If a 2-lane turning roadway is merged into one lane before the entrance 
terminal at the through roadway, what, in your opinion, is the most 
desirable length and taper ratio to be used? If your answer requires 
qualification please note the qualifications under comments. 







Merging length 


ft. 








Taper 


ratio 


:1 




Merging 


Length 


No. 


Responding 


Percent 


Cumulative % 


(ft. 


) 










200 






1 


5.5 


5.5 


400 






1 


5.5 


11.0 


600 






8 


44.5 


55.5 


800 






2 


11.1 


66.6 


1000 






3 


16.7 


83.3 


1200 






3 


16.7 


100.0 



Total 



18 



100.0 



Taper Ratio 


No. Responding 


Percent 


Cumulative % 


X:l 










20 


1 




4.8 


4.8 


35 


1 




4.8 


9.6 


40 


1 




4.8 


14.4 


50 


13 




61.8 


76.2 


60 


1 




4.8 


81.0 


70 


1 




4.8 


85.8 


80 


1 




4.8 


90.6 


100 


2 




9.4 


100.0 



Total 



21 



100.0 



B-75 



Comments : 

1000 ft. and 80:1 taper; however, slightly lower criteria such as 
700 ft. or a 60:1 taper are functional. 

We would prefer an auxiliary lane 2000-2500 ft. along the mainline 
rather than the construction on the ramp. 

Personally, I do not prefer this method since it usually will occur 
on a curve and it is not expected by motorists, especially if preceded 
by a two-lane exit and the ramp proper is adequate for 60-70 mph speeds. 
I feel that taper ratios much greater than 50:1 are too flat to be 
noticed and are conducive to sideswipes. 

(12 X velocity) concept — 600 ft. response. 

35:1 or possibly shorter will escape provision along the ramp shoulder 
for drivers unable to merge in alotted distance. 

400 ft. and a 50:1 taper assuming a width reduction from 28 to 20 ft. 
and a design speed of 50 mph. 

Merging length depends on offset. 

1200 ft. and 100:1 is not a recommended procedure under conditions 
requiring full two-lane capacity. 

200 ft. and 20:1 — This configuration is somewhat undesirable: it 
results in the ramp being used for storage. Each case must be carefully 
examined to insure that traffic does not accumulate to the point that the 
other mainline is blocked. 

This (600 ft. and 50:1) is assuming 12 ft. wide lanes on the ramp. The 
taper ratio may be adjusted if the lanes are less than 12 ft. wide, but the 
taper length should not be less than 400 ft. long, and the entrance lane 
should not be less than 12 ft. wide or the width of the right-hand through 
lane, which ever is greater. 



B-76 



40. If a two-lane turning roadway is merged into one lane before the 
entrance terminal at the through roadway, which lane do you think 
should be dropped? 

Number Responding Percent 

Left Lane 7 33.3 

Right Lane 12 51.1 

Either Lane __2_ 9.6 

21 100 

Comments : 

Right Lane: 

Conforms to driver expectancy on ramps — my answer for a right side 
merge onto freeway — (a) is probably better for left side merges to 
freeway. 

Only for consistency. 

Wider shoulder and lower speeds are on right side 

Right lane should be dropped unless for some reason large majority 
of traffic is already in the right lane in which case left lane should 
be dropped. 

This question can probably be debated either way — prefer merging 
from right to left. 

Left lane merge may invite left lane vehicle to enter mainline before 
desired. 

Left-to-right merge is difficult. 

Easier to merge. 

Caution should be the burden of the right-hand or "slower" driver. 

Left Lane; 

Dropping the lane on the left will normally affect fewer vehicles. 
I believe this to be more desirable even though the vehicles on the left 
may be operating at a higher speed. 

Normally, less traffic will utilize the left lane of a two-lane exit 
roadway, which can effect a lane drop more readily. 

Assuming conditions other than capacity require two lanes, such as 
storage requirements or a truck climbing situation. 

Either Lane: 



When entering the through roadway from the right, then drop the right 
turning roadway . 



B-77 



Cor 



41. Figures 23 and 24 indicate two methods for merging a 2-lane entrance 
ramp. Place an x under the use category which characterizes the 
frequency with which your organization currently uses each method. 



-I*- 










\- 



Xl 



23 




AND Ris/iT Thru L Ane. 



24 









n 














iH 






Figure 






Iq 










0} 




(U 






Number 






< B-? 

O 
*J o 

0) t-( 


H to 

03 vo 

-3 ^ 




(3 vo 
0) I 

O ^-- 




B /-^ 

•H B-? 

4J m 
0) ro 
B 1 

O VO 
CO ^-' 




!2; 

•U <-v 




Total 


23 


Number 


Responding 


6 


4 




1 




3 




2 




16 




% 




37.5 


25, 





6. 


3 


18. 


8 


12. 


5 


100 


24 


Number 


Responding 


1 


1 




3 




2 




9 




16 




% 




6.3 


6. 


3 


18. 


8 


12. 


5 


56. 


3 


100 



Comments : 

I believe operation as in #24 is very undesirable. 

When a two-lane entrance ramp is brought into the mainline, the mainline 
should have an added lane ahead. Neither figure shows this. 

We tried #24 and it does not work. We do not believe double lane drops 
work and have proved the same on several locations. 

Figure 23 gives a much better alternative for an escale lane (shoulder") 
than 24. 

Does not seem practical to use two lane ramp on and not Increase the 
mainline . 



B-78 



Because most of the existing system is like Figure 24, in the Chicago 
Region, it prefers to continue this type of design. 

My personal preference is for figure 24. 



42. Using Figures 23 and 24 as a reference please indicate your opinion as 
to the minimum and desirable values for the following dimensions for 
a 2-lane entrance ramp. 




\f^a- \_U3 



'^=^ 



^ 



Mefiac 2 EnTPANZf Lancs 



23 




-Mtfi<i£ Ltrr £nTPANCt Lahi 

AND RliHT r-if^U LaHC 



24 



Dimensions 

A. Distance from nose to start of merge 

B. Length of first merge 

C. Length of parallel auxilliary value 

D. Length of second merge 
X. Convergence ratio 





Fig. 23 


Minimum 


"A" 




A (in feet) 


No. Responding 


Percent 


Cumulative % 





1 




9.1 


9.1 


200 


1 




9.1 


18.2 


280 


1 




9.1 


27.3 


400 


1 




9.1 


36.4 


450 


1 




9.1 


45.5 


500 


2 




18.2 


63.7 


600 


3 




27,2 


90.0 


900 


1 




9.1 


100.0 



Total 



11 



B-79 



Fie. 23 Desirable "A' 



A (in feet) 


No . Resp 


Dnding 


Percent 


Cumulative % 





1 




9.1 


9.1 


300 


1 




9.1 


18.2 


400 


1 




9.1 


27.3 


450 


1 




9.1 


36.4 


500 


2 




18.2 


54.6 


600 


1 




9.1 


63.7 


750 


2 




18.2 


81.9 


800 


1 




9.1 


91.0 


1000 


1 




9.1 


100.0 



Total 



11 



Fig. 23 Minimum "B" 



B (in feet) 


No. 


Responding 


Percent 


Cumulative % 


400 




1 


9.1 


9.1 


600 




6 


54.5 


63.6 


750 




1 


9.1 


72.7 


850 




1 


9.1 


81.8 


900 




1 


9.1 


90.9 


1000 




1 


9.1 


100.0 



Total 



11 



Figure 23 Desirable "B" 



B (in feet) 


No. 


Responding 


Percent 


Cumulative % 


600 




5 




45.4 


45.4 


840 




1 




9.1 


54.5 


900 




2 




18.2 


72.7 


1000 




1 




9.1 


81.8 


1200 




2 




18.2 


100.0 



Total 



11 



Figure 23 Minimum "C" 



C (in feet) 


No . Responding 


Percent 


Cumulative % 





2 


18.2 


18.2 


400 


1 


9.1 


27.3 


500 


1 


9.1 


36.4 


600 


1 


9.1 


45.5 


800 


1 


9.1 


54.6 


900 


1 


9.1 


63.7 


1300 


1 


9.1 


72.8 


1500 


1 


9.1 


81.9 


1600 


1 


9.1 


91.0 


2000 


1 


9.1 


100.0 



Total 



11 



B-80 



Fig. 23 Desirable "C" 



C (in feet) 


No. Responding 


Percent 


Cumulative % 


600 


1 


11.1 


11.1 


750 


1 


11.1 


22.2 


1000 


2 


22.2 


44.4 


1200 


1 


11.1 


55.5 


1600 


1 


11.1 


66.6 


2000 


1 


11.1 


77.7 


2500 


2 


22.2 


100.0 



Total 



Total 



Fig. 23 Minimum "D' 



D (in feet) 


No. 


Responding 


Percent 


Cumulative % 


400 




1 


9.1 


9.1 


600 




7 


63.6 


72.7 


750 




1 


9.1 


81.8 


850 




1 


9.1 


90.9 


1000 




1 


9.1 


100.0 



11 



Total 



Fig. 23 Desirable "D' 



D (in feet) 


No. 


Responding 


Percent 


Cumulative % 


600 




6 


54.5 


54.5 


840 




1 


9.1 


63.6 


900 




1 


9.1 


72.7 


1000 




1 


9.1 


81.8 


1200 




2 


18.2 


100.0 



11 



X;l 

15 
30 
50 
70 
80 

Total 



Fig, 23 Minimum "X 
No. Responding 



Percent 


Cumulative % 


9.1 


9.1 


9.1 


18.2 


63.6 


81.8 


9.1 


90.9 


9.1 


100.0 



11 



B-81 



Fig. 23 Desirable "X" 



X:l 


No. 


Respondlr 


^§ 


Percent 


Cumulative % 


15 




1 




9.1 


9.1 


50 




6 




54.5 


63.6 


60 




1 




9.1 


72.7 


70 




1 




9.1 


81.8 


100 




2 




18.2 


100.0 



Total 



11 



A (in feet) 

300 
400 
660 

Total 



Fig. 24 Minimum "A" 
No. Responding Percent 



50 
25 
25 



Cumulative % 

50 

75 

100 



A (in feet) 

400 
600 
660 

Total 



Fig. 24 Desirable "A" 
No. I^esponding Percent 



25 
50 
25 



Cumulative % 

25 

75 

100 



B (in feet) 

600 

690 

700 

1000 

Total 



Fig. 24 Minimum "B 
No. Responding 



Percent 


Cumulative % 


25 


25 


25 


50 


25 


75 


25 


100 



B (in feet) 

800 

900 

1200 

Total 



Fig. 24 Desirable "B" 
No. Responding Percent 



33.3 
33.3 
33.3 



Cumulative % 

33.3 

66.6 

100.0 



B-82 



C (In feet) 

300 
2000 

Total 



Fig. 24 Minimum "C" 
No. Responding Percent 



50 
50 



Cumulative % 

50.0 
100.0 



C (in feet) 

400 
1300 
2500 

Total 



Fig. 24 Desirable "C" 
No. Responding Percent 



25 
25 
50 



Cumulative % 

25 

50 

100 



D (in feet) 

600 

690 

700 

1000 

Total 



Fig. 24 Minimum "D 
No. Responding 



Percent 


Cumulative % 


25 


25 


25 


50 


25 


75 


25 


100 



D (in feet) 

900 
1200 

Total 



Fig. 24 Desirable "D" 
No. Responding Percent 



66.7 
33.3 



Cumulative % 

66.7 
100.0 



Total 



Fig. 24 Minimum "X" 



X:l 


No. 


Resp 


onding 


Percent 


Cumulative % 


50 




1 




25 


25 


57 




1 




25 


50 


60 




1 




25 


75 


70 




1 




25 


100 



B-83 



Fig. 24 Desirable "X 



X:l 


No. 


Responding 


Percent 


Cumulative % 


57 




1 




25 


25 


70 




1 




25 


50 


75 




1 




25 


75 


100 




1 




25 


100 



Total 



B-84 



43. Figures 25, 26 and 27 indicate three methods for merging two 2-lane 

roadways into one 3-lane roadway. Place an "x" under the use category 
which characterizes the frequency with which your organization cur- 
rently uses each method. 



® 



-^ 



MiRCt 2 Right Lanes 



M^ 



25 



® 




\t- 



■MCPdE. 2 CCNTCR LAflES 



\(^ 



26 



® 



t — ^ 




\F 



■Ml/^CE 2. L£FT LAN€S 



M^ 



27 







CO 








u 








cd 








(U 








^ 








> 








iH y~~. 






CO 


(U 








< ^S 


/•~\ 


/"^ 


Q; 


a 








O 


!>>B-S 


^^ 


e ^ 










■u o 


r-\ \r\ 


m 


•H B-« 


•u ^^ 








03 rH 


H a> 


C vo 


j-i m 


to &^ 








O 1 


« 1 


0) 1 


cu n 


O lO 








e vo 


3 >£) 


w vo 


g 1 


S 1 








iH o^ 


CO vo 


IH 0-) 


O vD 


t-i o 








<: ^ 


P ^ 


O '^ 


W ^-- 


<; ^ 


Total 


FIGURE 
















25 


No . Responding 


6 


4 


2 


4 


1 


17 




Percent 


35.3 


23.5 


11.8 


23.5 


5.9 


100 


26 


No . Responding 


1 


1 


2 


4 


9 


17 




Percent 


5.9 


5.9 


11.8 


23.5 


53.0 


100 


11 


No. Responding 








-1 


4 


12 


17 




Percent 








5.9 


23.5 


70.6 


100 


Comments 


, 















Fig. 26 will not work. We have used with some success but we believe 
that it enhances operation if lane is chopped at exit ramp if T distance is 
less than 1500'. (Texas) 



B-85 



44. Presented below are several statements concerning the use of weaving 
areas on major interchanges. Indicate your agreement or disagreement 
with each of these statements by encircling the appropriate word. 

Statement 1. In general, weaving areas in major interchanges should 
always be avoided. 



a. Strongly 
Agree 



b. Agree 



c. Disagree 



d. Disagree 
Strongly 



Numb er 
Responding 



Percent 



Strongly Agree 
Agree 
Disagree 
Disagree Strongly 



4 
13 

4 




21 



19.1 

61.8 

19.1 





100 



Statement 2. In general, weaving areas in major interchanges can be 
justified if adequate weaving length is provided. 



Strongly b. Agree c. 
Agree 


Disagree 


d. 


Disagree 
Strongly 




Number 
Responding 


Percent 






Stron]gly Agree 
Agree 
Disagree 
Disagree Strongly 


1 
13 

7 



4.8 
61.9 
33.3 









21 



100 



Statement 3. In general, weaving areas in major interchanges can be 
justified if all weaving occurs off the through roadway. 



a. Strongly 
Agree 



b. Agree 



Disagree 



Disagree 
Strongly 



Number 
Responding 



Percent 



Strongly Agree 3 

Agree 13 

Disagree 5 

Disagree Strongly 



14.3 

61.9 

23.8 





21 



100 



B-86 



45. (Deleted) 

46. The amount and quality of feedback design engineers get regarding opera- 
tion and safety of interchanges seems to vary considerably from organi- 
zation to organization. Please comment below on the ways in which you 
think feedback could be provided so that the information would be useful 
on subsequent designs. 



Research studies. 
Operational reports 

The Maintenance Division could issue periodic reports on operational 
problems and supply copies to the Design Subdivision. (New York) 

Our Department is on the "non-centralize" system and all of the inter- 
change design is accomplished in the district offices and approved by the 
Austin Office. Therefore, we have very good feedback because our design 
engineers live with the problems they create. 

Periodic written reports from operations should be made to design on 
performance of various design features. Standardization of accident report- 
ing would be very desirable so that correlation with design features would 
be possible. 

1) Accident data in a better form (summarized so that designer could 
more readily get the overall picture) . 

2) Actual traffic counts taken by experienced men who could make deter- 
minations, estimations as to the level of service, and other factors so 
designers would have a better feel or understanding what 600 vph or 1000 vph 
would mean. (films would be great.) 

All major interchanges constructed should be automatically subjected to a 
1-2 year special study of operation and safety aspects. The data should be 
sent to the Central Headquarters, analyzed, and the results used to influence 
subsequent designs. 

Accident reports with copies of investigation results, if any, should 
be routinely transmitted to design for the first few years after opening 
a new facility. Copies of comments from public, police, maintenance should 
also be sent to design. Traffic studies should be made at frequent inter- 
vals and made available to Design. 

1) Deliberate program of feedback from maintenance and safety offices 
regarding outstanding deficiencies in operation and why. Also report out- 
standing designs. 

2) Nationwide reporting of same as above. 

Maintenance men 

Traffic operations division 

Fire departments 

Local and state police 

Trucking associations 

Traffic safety division 



B-87 



Bulletins published through one agency — AASHO or FHWA. Too many 
agencies or firms have been publishing safety data, depending on what is 
being sold or published. 

Engineers drive completed facility 

1) During and upon completion of construction, receipt of as-built 
drawings which specify geometric changes because of constructability problems. 

2) Visit and record driver behavior at interchange on first day of 
operations. 

3) During operations, feedback concerning accident locations, special 
sign requirements and experienced operational data regarding speeds, volumes, 
and geometries which can be correlated to original or as-built design. 

Experienced traffic operations engineers should be an integral part 
of the design team. 

1) Closer communication between design, maintenance, and enforcement 
personnel. 

2) Designers spending more time in field observing traffic operation. 

Multi-disciplinary diagnostic team to evaluate projects (see Research 
Report 606-8 copy enclosed) . 



B-88 



APPENDIX C 
WORKSHOP ATTENDEES & AGENDA 



C-i 



^ 



APPENDIX C: WORKSHOP ATTENDEES & AGENDA 

Invitations for the two workshops held in August were sent to 
selected state highway engineers. Federal Highway Administration 
officials, private consulting engineers, and academic-research engineers. 
Representative of the top highway designers and design policy makers 
in the country, the attendees are identified in Table C-1. 

Held over a period of three days, each workshop consisted of 
13 discussion sessions approximately 90 minutes in length » (See Table C-2.) 
The general session format consisted of the following: 

. An introduction by one of the project personnel, acting as 

discussion leader. 
. A short summary of the information derived from the litera- 
ture, interviews with highway department personnel, and 
completed pre-workshop questionnaires » 
. Presentation of a set of prepared questions, and encourage- 
ment of expressions of opinions and relating of pertinent 
experience by the workshop participants = In addition, a rea- 
sonable amount of "open" discussion in the general topic 
area was encouraged « 
. Following the discussions, distribution of a prepared 
"session questionnaire," The participants were asked to 
provide written answers or opinions to specific questions. 



C-1 



TABLE C-1. Workshop Attendees 
First Workshop (August 1-3, 1972) 



Biggs, Raymond G. 

Project Superxisor (Houston) 

Texas Dept. of Transportation 

Coble, Kenneth L. 
Consulting Engineer 
Sverdrup & Parcel 

Fields, Marvin 

Field Engineer, Bureau of Traffic 

Ohio Department of Highways 

Foy, Robert A. 

Chief Engineer, Design Division 

Wilbur Smith & Associates 

Gray C. William 

Design Development Engineer 

Ohio Department of Highways 

Hall, Parker L. 

Assistant Engineer of Design 

California Division of Highways 

Hess, Joseph W. 

Office of Research (FHRS) 

Federal Highway Administration 

Housworth, Jack L. 
Supervising Design Engineer 
Texas Dept. of Transportation 



Hurd, Fred W. 

Professor, Civil Engineering 

The Penna. State University 

Kenyon, Alan D. 

Associate Civil Engineer 

New York Dept. of Transportation 

Link, James 

Office of Development 

Federal Highway Administration 

McCausland, Walter 

Design Engineer 

Federal Highway Administration 

McCoy, William D. 

Assistant State Highway Urban Engineer 

Georgia Highway Division 

Mueser, Robert R. 

Deputy Chief Highway Engineer 

Penna. Department of Transportation 

Nemeth, Dr. Zoltan 

Civil Engineering Department 

Ohio State University 

Stockfisch, Charles R. 
Office of Research (FHRS) 
Federal Highway Administration 



n 



Second Workshop (August 22-24, 1972) 



Alexander, Dr. Gerson J. 
Office of Traffic Operations 
Federal Highway Administration 

Byington, Stanley R. 

Office of Research 

Federal Highway Administration 

Churchill, Robert R. 

Deputy Design Engineer 

Florida Dept. of Transportation 



Lins, William F. 

Chief, Bureau of Highway Design 

Maryland Dept. of Transportation 

Loutzenheiser, Donald W. 
Director, Office of Engineering 
Federal Highway Administration 

McCausland, Walter 

Design Engineer 

Federal Highway Administration 



c-2 



I 



TABLE C-1 (Continued) 



Ebersole, Glenn 

Traffic Research Engineer 

Pennsylvania Dept. of Trans. 

Foster, W. M. 

Assistant Director 

Washington State Highway Dept. 

Gazda, Andrew J. 

Engineer of Geometric Design 

Illinois Dept. of Transportation 

Glenn on. Dr. John C. 

Manager, Traffic Safety Center 

Midwest Research Institute 

Hofmann, Frederick J. 
Senior Highway Engineer 
Edwards and Kelcey, Inc. 

Huckins, Edgar W. 

Assistant Highway Design Engineer 

New Hampshire Dept. of Pub. Wks. 

Lee, Bumjung 

Research Associate 

Polytechnic Institute of Brooklyn 



Pilkington, George 
Office of Research 
Federal Highway Administration 

Randich, Gene M. 

Vice-President 

DeleuWj Gather Organization 

Ricker , Edmund 

Chief - Highway Safety Group 

Pennsylvania Dept. of Transportation 

Sigal, Andre H. 

Associate Civil Engineer 

NoYo State Dept. of Transportation 

Stockfisch, Charles Ro 

Office of Research 

Federal Highway Administration 

Tar agin, A. 

Office of Traffic Operations 

Federal Highway Administration 



C-3 



TABLE C-2. Workshop Agenda 
First Day 

Introduction, by James I. Taylor 

Discussion on Standardization; Classification; Adaptability led by 
Richard A. 01s en 

Discussion on Configuration Evolution Led by Robert Hos tetter 

Discussion on Design Sequence; Checklists led by John Hayward 

Second Day 

Discussion on Trade-offs; Cost-effectiveness; Level of Design 
led by James I. Taylor 

Discussion on Visibility Analyses, Driver Perception, design 
led by Richard A. Olsen 

Discussion on Exits led by Ronald J. Slavecki 

Discussion on Entrances led by Robert Hostetter 

Discussion on, New Designs led by John C, Hayward 

Third Day 

Discussion on Lane Drops; Lane Balance led by Ronald J. Slavecki 

Discussion on Route Continuity; Ramp Arrangements led by John Hayward 

Discussion on Local Access; Freeway Control; Bus Lanes led by 
Robert Hostetter 

Conclusion by James I. Taylor 



C-4 



I 



APPENDIX D 
SELECTED POST-SESSION QUESTIONNAIRE RESULTS 



D-i 



APPENDIX D: SELECTED POST-SESSION QUESTIONNAIRE RESULTS 

This appendix contains the results of the post-session question- 
naires which were distributed to all workshop attendees immediately 
following the workshop discussion sessions. The questionnaires were 
distributed so that the workshop participants could provide written 
opinions or statements concerning topics which had been discussed dur- 
ing the preceeding workshop session. These written responses were 
intended to reinforce the findings and trends noted in the pre-workshop 
questionnaire and to distill the session discussions, as well as 
draw out opinions from some of the less vocal workshop attendees = 

The post-session questionnaire results for some of the sessions 
have been deleted from this appendix since they are directly contained 
within the text of the preceeding report. Results from the' following 
workshop sessions are tabulated in this appendix in the following 
order: 

1) Exits; 

2) Two-lane entrances; 

3) Lane drops and lane balance; 

4) Route continuity. 

Each question will be reproduced as it was presented to the partici- 
pants followed by a summary of the answers received. Except for 
those questions which are indicated, the answer matrix will include 
the results of both workshops combined c Comments en each question 
were encouraged and many of those received will be presented following 
the answer matrix. 



D-1 



EXITS 



1. Cloverleafs are not adaptable for freeway-to-freeway interchanges, 
except possibly in rural areas where turning volumes are relatively 
low, and then only when the design includes collector-distributor 
(C-D) roads. (Agree or Disagree?) 





Stron 
Agree 


giy 


Agree 


Disagree 


St 
Di 


rongly 
sagree 


Tota 


Design 

Operations 

Research 

Total 


6 

1 
1 

8 




9 
1 

1 

11 


3 
3 
4 

10 




1 

1 

2 


19 

5 
7 

31 


Comments: 

















The design can be made without C-D roads if the design is such that 
C-D roads can be provided in the future when volumes and weaving 
indicate need. 

Cloverleafs are quite acceptable in low volume rural areas without 
C-D roads and where a route is not turning. Even in some suburban 
areas where 50 mph speeds are used and weaving can be accommodated 
at level of service D and space is available, cloverleafs are 
acceptable if space is available and the level of service on the 
mainline is D or less. C-D roads on cloverleafs are not always 
cost-effective. 

Although the construction cost is higher, directional interchanges 
are the only type we should build because they will be adequate for 
more years. 

Only because of cost do I believe that cloverleafs should ever be 
used. 

Weaving sections should not be placed on the mainline. 

From a safety standpoint, cloverleafs are usually no problem if the 
loop is on an upgrade and clearly visible to the driver. 

Loops might be used for turns of minor volumes. 

Generally I agree; however, we feel that the C-D roads are the con- 
trols. If_ you can design C-D roads of adequate length to handle 
the weaving volumes, we see no reason to abandon the cloverleaf 
design as an alternate in any location. 

They can be used successfully without C-D roads, most likely where 
there are not directional interchanges in the area and where volumes 
are low. 



D-2 



When through volumes are also very low, the weaving adjacent to 
the main lanes may be accetable. 

Short weaves of < 1,800 ft. should not be permitted on any freeway 
mainline. 

It depends on several factors; e.g., the balance of turning volumes. 
A C-D is not always a good answer. 

Where turning volumes are low, a C-D should not be needed. 

In my opinion, if low volumes are prevalent, an adequately long 
deceleration lane for achieving an appropriate speed reduction 
appears satisfactory. 

It is not quite practical to spend a million dollars to save a 
few injuries when there are pressing social problems which also 
need funds. 

In rural ai^eas C-D roads may not be required. But this would seem 
to be an exception rather than a rule. If C-D roads are not used, 
the designer should be made to justify it just as strongly as if 
he were proposing a very complex design. 



2. The disfavor with which left-hand exits are held by engineers stems 
more from subjective speculation than from the results of factual, 
objective studies. (Agree or Disagree?) 





Strongly 

Agree 


Agree 


Disagree 


Strongly 
Disagree 


Tota 


Design 

Operations 

Research 

Total 



1 


1 


3 
2 
4 

9 


12 

1 

15 


3 
1 
1 

5 


18 
5 
7 

30 


Comments : 













Left-hand exits are adequate with a two-lane roadway, if they are 
signed and lighted properly. 

I have formed my opinion of disfavor with left exits from the com- 
bination of having read results of studies of the subject, from 
my own experience with many such designs, and from knowing the 
opinions of other highway engineers. Any single published study 
I know of would not by itself prove the point. 

I think facts are available. The definition of a left-hand exit 
needs clarification. The following is bad: 



1,000 vph' 



'■ ^Hltlftt 



6,000 vph 



D-3 



There are several reports and a Congressional Hearing that indi- 
cate left-hand exits are not desirable. 

My state collects some accident data, and I am sure that it will 
show that left exits have high accident experience. 

All variables have not yet been controlled in existing research. 

Both factors are involved; however, study results appear to sub- 
stantially predominate. 

Although not many states, agencies, etc. have conducted their 
own research since most do not have enough situations to warrant 
statistical significance, they probably have, as Illinois, con- 
tributed funds toward regional or national studies by professional 
organizations to research the problem and provide results. One 
such study is the Illinois Cooperative Highway Research Project, 
IHR-61. I do not feel these can be overlooked just because they 
are not conducted by the utilizing agency. 

I must admit that good documentary evidence is not what it should 
be, but we have good, practical, visual evidence that left-hand 
exits are poor. We have designed and built a number of them. 
Without exception , traffic conflicts can readily be observed at 
each, far out of proportion to the volumes. 

I do not really know. I would like to think that some operational 
research is available. 

It is based on known operational problems. 

Local accident data has led many teams to this conclusion. There 
has been no big national "pull-together" report. 

The disfavor is based mostly on sad experience (operations and 
safety) with left-hand off-ramps. 

I do not believe that the speed difference between lanes on most 
freeways is so large that an adequate deceleration lane will not 
do the job. Perhaps the only problem that arises is where an 
entrance ramp is so close to the exit ramp that commercial vehicles 
who entered at the entrance ramp have inadequate space (longitudinal) 
to merge left to exit. 

There are a few studies related to left-hand ramps. 

Accident statistics will usually reflect the poor design in 
choosing to use a left-hand exit. 

Left entrances have been found to be worse than left exits from a 
safety standpoints 



D-4 



3. If a left-hand exit must be used, a parallel- type left lane should 
be added to the mainline to remove exiting traffic from the high- 
speed through lane. (Always or Never?) 

Always Usually Sometimes Rarely Never Total 



Design 


10 


6 





1 





17 


Operations 


2 





1 








3 


Research 


2 


3 





1 





6 



Total 14 9 1 2 26 
Comments : 

This is Ohio's standard for either left or right exits. 

The parallel type deceleration lane is rarely used in Texas. 

It is highly desirable to facilitate advance signing and provide 
vehicle orientation. In addition, this provides a deceleration 
lane out of the high-speed lane of the through facility. 

I want to say "always," but I cannot be quite that positive! How- 
ever, the more opportunity for early decision-making on left-hand 
exits the better. This feature should be included as a "given" 
to be dele;ced only in rare or the most unusual situations and 
then only if a good escape zone can be included ahead of the 
rampr 

It depends some on mainline alignments. It needs special treat- 
ment, and an added lane is one part. Special advance signs are 

also needed. 

You need adequate distance for signing, preferably overhead. 

Sometimes, depending on volumes, percent of heavy trucks, and 
topography (vertical alignment) . 

To prevent a reduction of speed on the mainline. 

Especially true if the mainline is curving to the right. A parallel 
lane will help reduce the speed differential between successive 
vehicles in the high-speed lane. 

4. Should a. federal standard be adopted prohibiting the use of left- 
hand exits at major interchanges? Why or why not? 

Yes No Total 



Design 


2 


15 


17 


Operations 


1 


3 


4 


Research 


2 


5 


7 



Total 5 23 28 

D-5 



Yes 

They are accident prone. 

Left-hand exits are inherently hazardous , 

On Interstate and maybe U.S. Routes. 

If it allows leeway in cases where there is no other feasible 
alternative. 



No 



If properly designed, they are safe. 

The standard should not prohibit, but should make their use 
very restricted. There are cases where all right-side exits 
would not be feasiblec 

The FHWA will not approve left-hand exits unless they are major 

splits. 

There is always she unusual case where one may be necessary 
although not desirable » 

There are instances when they must be provided. This should be 
left to the states. 

No, because there is little difference to the driver between a 
directional split of a freeway and a freeway-to-freeway left 
ramp, 

Undex cert^iin conditions left-hand exits may be warranted. 

With good signing it does not have to be a problem. At complex 
interchanges it may be a very economical solution (less ROW) . 

Such a pclicy would not be desirable. Although the left-hand 
exit is not generally preferred, it can be successfully used 
when properly designed and signed. 

No, because there is no standard which can be substituted for 
engineering judgment. 

These exits are poor. They should be used only as a last resort. 
Nevertheless J they are a tool or a method that can be used, if 
well designed, with at least satisfactory results where the 
tcpogxaphy may be such that nothing else would be tolerated. 
Federal standards are totally inflexible and do not recognize 
special conditions -. 

Extreme ccnditions may dictate the use of the left exit. 

They may have co be used. They aan be designed given the right 

situa*:icn. 

D-6 



There may be certain cases where this is the only practical solu- 
tion to the problem. 

They are second choice forms, but may be better than none at all. 
We can design, sign, and mark them to work well, but in earlier 
cases we did not do so. 

I do not consider federal standards to be the proper method to 
obtain good designs. There are cases in major interchanges where 
the left exit may be the most desirable solution. 

The designer should have a wide degree of flexibility in design 
approach — weighing cost, site, etc. 

No, because a general prohibition may create problems in some 
cases when a left-hand exit is the only way to solve the problem. 

There are exceptions in which left exits can be justified — 
depending on volumes; land available for building the interchange, 
etc. 

Other 

If this includes a major fork, no; if not, then yes. 

5. Direct taper off-ramps are, in general, superior to parallel- 
lane off-ramps. (Agree or Disagree?) 

Strongly Strongly 

Agree Agree Disagree Disagree Total 

Design 4 9 3 3 19 

Operations 3 115 
Research 3 3 17 

Tocal 4 15 7 5 31 

Comments : 

In the case of an accident on the off-ramp the parallel lane is 
good for safe storage (the paved shoulder also) . 

I believe that one type of exit should not be used exclusively. 
The tapered exit is advantageous under normal traffic conditions. 
Where excessive ramp volumes may occur, the parallel lane should 
be used. 

Parallel-type deceleration lanes are superior to tapered-types! 

I strongly agree for a single lane and low volumes. This is not 
true for two-lane branches, and not necessarily true for high 
volume single exits. 

One is as good as another, though we use the direct taper, 

D-7 



The parallel, if designed properly, will have all the advantages 
of the direct taper as well as its own advantages. 

It would depend on the location with relation to cross streets 
which may cause capacity problems requiring storage adjacent to 
the freeway. 

The parallel lane has merit where the exit condition is less than 
desirable for various possible reasons, 

I believe the results of the 1960 AA.SHO Special Study indicate 
that it is generally true that direct taper deceleration lanes 
are superior. 

Generally I agree since many motorists mistake the parallel lane 
for an added lane. I agree only if the direct taper has an ade- 
quate escape zone, however. 

It more nearly approximates driver activity » 

Best for low voltime rural areas — driver use pattern verifies this. 
It is the natural turning motion. 

It must be kept in mind that certain factors ^ such as curves 
creating poor sight distance, could dictate parallel-lane off- 
ramps. 

Either can be designed to work well. Tapered is in favor for 
non-major interchanges, but views are divided for use at major 
ones. It is not correct to say that a tapered type is "superior" 
for all cases. 

There are special cases, such as exits on the outside of horizontal 
curves, where the use of the parallel lane is preferable. 

On high volume highways, I believe that long parallel lanes are 
safer and less restrictive to possible capacity problems. 

Only when there is no tangential off- ramp. 

New Jersey Turnpike experience indicates that parallel lanes are 
good. 

Tapers, in my opinion, provide less direction to drivers — depends 
on striping. How is the driver guided along the tapered section? 
I believe the only problem with acceleration and deceleration lanes 
is that we have not taught drivers what they are for and how to 
use them. 

Getting exiting traffic away from the through lane is desirable. 

Both can be used depending on the situation. If the road curves 
to the left, a taper may be used rather than a parallel lane. In 
urban areas where density is usually high, the parallel lane can 
increase capacity at the ramp. Parallel lanes can operate as a 
taper when traffic demands drop off, 

D-8 



so 







r-H 








(0 








4-1 


vO 


/"-S 




O 


iH 


& 




H 




bO O 








C rH 








•H QJ 








CJ W 








»-i 








3 4J 


iH 






H Xi 


0) 






M 


iH 






CU iH 


iH 


0) 


O 


« 


C 


iH 


Ti *-> 


^ 


n> 




iH 


t-I 




C -W 


Ah 






•H M-l 








;g^ 








« 


4J 






w 


o 


H 






<u 


0) 


CO 




u 


Pi 






•H 


cd 






O 


H 





CS| 



cn| 00 







O 

H 


CO 






u 






bO 






o 


iH 




o 


0) 




OJ 


rH 


OJ 


c 


OJ 


C 


tH 


>-l 


CO 


.H 


CO 


i-I 


c: 


(U 





lO 



1^ 

CM 



LO 



o 


U 




(U 


0> 


o 


M 


a 


^ 


-ri 


CO 





lo 



Q H 



I 



O 
H 



rt 






0) 






M 






c 






CO 






E-i 






<U 


rH 


OJ 


a 


CO 


g 


H 


^ 


^4 


C 


CO 




•H 


Ph 







O H 




(U O 




H ft 




•H CO 




O H 


OT 




H 




M 




X! 




W 





CM 



m 



fo 







rH 








CO 


VO 






u 


rH 






O 




«J 




H 




05 








<0 








u 








u 








^ 


(H 






0) 


<U 






> 


iH 


(U 




o 


iH 


fl 


CI 




as 


CO 


«H 


4J 


^ 


i-J 




°H 


ifl 






X 


Ph 






w 








6 


w 








u 


^ 






0) 


0) 


en 




!M 


P. 






"H 


CO 






P 


H 





CsJ 



«n 



CM 





CO 








c 








o 


^ 






•H 


o 




c 


+J 


v< 




60 


CO 


CO 


rH 


•H 


h 


0) 


CO 


CO 


(U 


CO 


4-1 


0} 


p< 


0) 


o 


o 


o 


Pi 


H 





03 








c 








O 


^ 






•rl 


o 




a 


4-1 


Kl 




60 


CO 


CO 


rH 


"H 


n 


(U 


CO 


CO 


0) 


CO 


«J 


0) 


p. 


<^ 


o 


« 


o 


rt 


H 



D-9 



a 

•H 

c: 
u 

H 
0) 



•S «J fi!; 



o 



0) 
H (U 

<0 



vo 



pq 



4J 

•rl tO 
P H 



en 



CO 



>* 




CO 


CM 




•u 




0) 


O 




T3 


H 




c4 






U 






00 






a 






& 






o 






o 


iH 




0) 


Q) 




c 


f-i 0) 


O 


-H 


iH fl 


CM 


.H 


CO CO 




c: 


>^ h^ 




•H 


CO 




^ 


Ph 




• 


U VI 
0) 01 


-* 




■H CO 



CM 



cr> 



CM 



cn 



lO 



CM 



Q H 



CO 

EH 

a 



I 



CO 

O tH 
H 



iH (U 

CO CO 

H !-:» 
CO 



o 


Kl 


cu 


(U 


Vl 


fX 


•H 


CO 



lo 



O H 



vH CO 



1 









CO 




>* 






iJ 


«* 


CM 


4J 
CO 
0) 

o 

0) 

> 


0) 


o 


M 




o 


^ 


<i) 




lO 




^ 


S 


.n 


iH 


■W' 


aj 


M 




•H 


Vi 


^J 






X 


CO 








w 


Ph 








o 


4J 

O 
(1) 


VI 

0) 





CM 



fO 



m 



CM 



CO 






C 








o 


js 






•H 


o 




(3 


U 


u 




bO 


CO 


CO 


rH 


•H 


Vi 


0) 


CO 


to 


0) 


0) 


4J 


(U 


cu 


«i 


o 


« 


o 


Pi 


H 





CO 








C 








o 


^ 






•H 


o 




c 


•U 


VI 




60 


CO 


CO 


r-i 


»H 


M 


0) 


CO 


CO 


Q) 


CO 


4J 


(U 


O. 


«J 


O 


« 


O 


Pd 


H 



D-10 



7. Assuming a highway design speed of 70 mph and an exit ramp speed of 
50 mph, the "Blue Book" (1965a) suggests a deceleration lane of 350 
feet (of which 300 feet is the length of the taper) , 

a. Do you feel this length is adequate? 

Yes No Total 



Design 


7 


10 


17 


Operations 


1 


4 


5 


Research 


2 


4 


6 



Total 10 18 28 

b. If you feel the length is inadequate, what length would you 
suggest? 

Minimum Length (f t . ) Number Responding 



335 


1 


400 


1 


450 


1 


500 


6 


600 


2 


700 


2 


800 


3 


1,000 


1 



Total 17 
Desirable Length (ft,) Number Responding 



650 


2 


700 


1 


800 


2 


1,000 


5 


1,200 


2 


2,000 


1 



Total 13 



8. In general, a single exit is superior to a double exit (two succes- 
sive exits) in terms of driver comfort and confidence in a semi- 
directional interchange. (Always or Never?) 





Almost 




Occa- 


Rarely or 






Always 


Usually 


sionally 


Never 


Tota 


Design 


8 


8 


2 


1 


19 


Operations 


1 


1 


1 





3 


Research 


_4 


_3 





_0 


7 


Total 


13 


12 


3 


1 


29 



D-11 



Has enough research been done, or do you have enough results from 
your own experience that you are confident you understand the 
benefits of a single or a double exit on a semi-directional inter- 
change? (Confident or Unsure?) 

Totally Reasonably Some 

Confident Confident Doubt Unsure Total 

Design 4 12 1 1 18 

Operations 1 1114 

Research 15 17 

Total 5 14 7 3 29 



D-12 



TWO-LANE ENTRANCES 



1. What are the primary operational problems you have observed on two- 
lane entrance ramps? (Comment) 

Inadequate signing, lighting, and overhead signal control. These 
three elements can increase safety and capacity o The cost is 
low, but present policy prevents it. 

The breakdown of traffic flow in at least one lane occurs at a 
critical decision-making point of traffic flow. 

The merging length is too short and there is usually a lack of 
extra lane length beyond the entrance nose before merging to 
the minimum number of through lanes. 

Capacity, lane orientation, striping difficulties. 

Insufficient gaps in the right lane of the through roadway during 
peaks. 

The design of entrance and exit terminals. When designed as a 
major fork, they have been much more effective and provide 
smoother operations. 

Merging where separate lanes cannot be carried ahead „ 

They function as only one lane due to inadequate merge distance. 
There are problems if one lane is not continued » 

Compounded merging. 

Insufficient lanes going ahead. If L = the nvtmber of mainline 

lanes and L = the number of ramp lanes: 
R 

L + L„ to L + L„ or L + L„ - 1 lane = 0,K. 
m R m R — m R 

L + L„ to L + L„ - 2 lanes = insufficient lanes, 
m R m R 

The right-hand ramp lane traffic attempting to merge with the 
mainline too early; i.e., moving across the left-hand ramp line 
immediately past the gore area. 

Excessive lane changing; failure to use both lanes by "country" 
drivers; operating too slowly in the inaer lane must merge, 
(The above operational experience is based on older designs 
where the inner lanes were merged and the overall design was 
too short.) 

We have two-lane directions merging with two lanes and have no 
problems. 



D-13 



Merging is a problem if an additional lane is not provided, the 
length of acceleration lanes are not long enough, or acceleration 
lanes are not dropped as two separate lanes but as one contin- 
uous merge. 

X have heard of merge backup and accident problems. 

When an additional lane is not added to the freeway, merging 
creates congestion when the main lanes are near capacity. 

Erratic driver patterns at or just before the nose; uncertainty 
in desired position, 

Confusion regarding lane assignments; sideswipe collisions; outer 
lane drivers trapped on the acceleration lane. 

There is a problem with weaves, unless the exit ramp is well 
down the road. 

Driver confusion. 

a. Not functional in the sense that drivers will not use it as 
a two-lane facility, unless they are operating at capacity, 
and then the merging problem is paramount, 

b. Their safety record is not very good. 

Unbalanced use of lanes; merging difficulties for inner lane 
traffic resulting in vehicles stopping on a through-type facility* 



2, Given a two-lane entrance ramp where only one freeway lane is added, 
please indicate your preference for the merging lane configurations 
shown in Figure D-^l by giving the most desirable configuration a 1 
and the least desirable a 3. 

Average Rank (Both Workshops Combined) 

ABC 
Inner Lane Outer Lane Non-Compulsory 
Merged Merged Merge 

Design 2.4 1.6 2.1 

Operations 2.8 1,5 1.8 

Research 2A lj5_ 2^ 

Total 2.4 1.5 2,2 



D-14 



09 

a 
o 

u 
cd 

3 





a 




o 


0) 


o 


M 




M 


0) 


.2 


a 


S 


cd 




h7 


>. 




u 


bO 


o 


a 


m 


Ti 


tH 


bO 


3 


M 


& 


OX 


g 


a 


O 




o 


«• 


r 




c 





o 


CD 


^ 


j^ 



u 




0) 

o 
u 




2 

3 
bO 
•H 

pC4 



D-15 



3. Please comment on the factors which determined your ranking of the 
merging configurations. 

We need lighting, good signing, lane marking, and public educa- 
tion to make these things work. A stranger, when traffic is 
heavy, follows the car in front. Good traffic control devices 
take care of the stranger during low volume periods and at night. 

In my view, the merging process is a merging of ramp lanes and 

should be accomplished as a ramp function so that conflicts 

and interruptions of flow are not created for the mainline lanes. 

1. Flexibility of traffic volume ranges on each roadway, 

2. When requiring the outer lane (lane B) to merge, it is 
believed to require more vehicles to change lanes, especially 
during lower-volume, non-peak hours. 

It is best to lose the lower-speed outside lane. It is more 
consistent with the merge left concept. Drivers expect this. 
"A" creates a "squeeze play," 

New York State will not build a ramp with the left lane merged, 

"A" traps the unfamiliar driver with a merge left or right with 
no place to go. 

I feel a person should always be given an escape lane; therefore, 
I prefer the outer lane merge. 

B and C are basically the same. 

A — the squeeze merge is unsatisfactory. 

I would prefer the addition of two lanes with a right-side lane 

drop after the maneuver. It would change from a merge to a lane 

change . 

Eliminate the squeeze. 

Elemental merge. 

Keep intersectional maneuver simple. 

One decision at a time. 

Provide more distance for the lane change. 

B - Clear case - Lane B must merge, but it has an escape route 

(shoulder) . 
C - Not so clear - Lanes A and B merge into one lane on a 

"first come-first served" basis. 
A - Drivers in lane A have no escape route. 

The outer lane merge has better driver visibility, lower operat- 
ing speeds, and normally more available recovery area than 
either of the other designs. The non-compulsory merge is indeci- 
sive in its instructions, leading to gross confusion. 

D-16 



1 



The outer lane, being of generally lower speed, has the best 
opportunity with the least conflict, and also has more escape 
options. 

Reason for ranking (C,l; B,2; A, 3) include: 

1. My observation of the inner lane merge design (our former 

policy) . 
2e When turbulence occurs near peak hours, the inner merge 

accidents have always been high number multiple-types and 

mere serious because of no escape route. 

3. The outer lane merge has a full paved shoulder for escape 
or auxiliary lane usage, and the accident experience 
usually involves only minor rear-end collisions, 

4. It is my recollection that the special ITE study of 
this matter concluded that either design is satis- 
factory if uniformity is used within the locate. 



The shoulder provides an escape value, and there are less poten- 
tial vehicle conflicts if the outer lane is merged. 

The public expects the right or outer lane to merge. It has 
worked better in Maryland. It is simpler to sign, and the 
shoulder can be used for an escape lane. If an accident occurs, 
it will involve less cars at lower speed. 

Do not force the inner lane driver into a squeeze by ending 
his lane where he is trying to pull into the through lane. 

The inner lane merge does not permit any area for escape if a 
gap is net available. The outer lane traffic always has the 
shoulder as an escape area which is a safety factor- 

I do not like pinching out an interior lane. 

Desirability of having an escape shoulder on the right. 

Normal merge is to the left. 

Prevent a driver from having to decide whether to merge left or 

right. 

The effects on the through system (preferred configuration A) 

Safety 

Delay at the ramp 

By dropping the outer lane the driver has some flexibility due to 
the availability of the shoulder as an escape. The inner lane 
merge gives the driver no alternative. The non-compulsory merge 
forces the driver to make a decision that could be made for him 
by designating which lane will drop. 



D-17 



LANE DROPS AND LANE BALANCE 



In the pre-workshop questionnaire. Figure 14 was asserted to 
be the configuration most often used by the organizations 
and agencies of the respondents to reduce the number of main- 
line lanes from three to two. Please rank these three con- 
figurations in the order you personally prefer to employ them. 



JC 



■3 Lanes 



3 LaneSjt 



13 



Drop Lane 

BerOND fNTCFCHANee 



^Lj 



2 Lan£S 



•7 




^ Drop Ri6ht Lane 



-3 Lanes 



Drop Left Lane ■ 



i Lanes - 




The Number of Experts Assigning the Specified Rank 



Design 



Relative 
Ranking 

1 
2 
3 



Figure 
13 

11 
5 




Figure 
14 

6 
10 

1 



Figure 
15 



1 
16 



1 



Operations 



Research 



D-18 



I 



In the pre-workshop questionnaire. Figure 18 was asserted to 
be the configuration most often used by the organization and 
agencies of the respondents to reduce the number of mainline 
lanes from four to three following a one-lane exit. Please 
rank these three configurations in the order you personally 
prefer to employ them. 



j-4 Lanes 


4 Lanes^ 




= == = = = = = = == = = = = = = = 


17 


/"Tr~^^~P^==:5-,_ Drop Lane 

Beyond Interchange 


^4 LA/ves 


3 LANE%-f 


18 


■ — — brrrr:::~~-____^ '^ — Di?op Right Lane 


^4 Lanls 


Drop LCFT Lane -^ j^^^^^_^ 


19 


~~^=^»-- 



The Number of Experts Assigning the Specified Rank 





Relative 
Ranking 


Figure 
17 


Figure 
18 


Figure 
19 


Design 


1 
2 
3 


12 
5 



6 
10 

1 




1 
16 


Operations 


1 
2 
3 


2 
3 



3 
2 





5 


Research 


1 
2 
3 


3 
2 



2 
3 





5 



D-19 



In the pre-workshop questionnaire, Figure 21 was asserted to 
be the configuration most often used by the organizations and 
agencierof 'the respondents to reduce the number of mainline 
lanes from four to three following a two-lane exit. Please 
rank these three configurations in the order that you person- 
ally prefer to employ them. 



,^'1 LANrs 












4 L AN£S-p, 


= = ~\'^ = = 


= = 






=: 


— = 




20 




'^^^'^^^^ 


^ 


^- 


Drop Lane 

BtroND iNTEKMfiHCe 


y'J LAIiEi 












■3LAHCSy 


^_^-feE = = = 


= = 


5E 




= 


= = 




21 






^ ^^^Ss ^ 


?= 


^ — D^op Right L mi£ 


^4 Laucs 






Dkop L £Fr 


LAne^ 


3 lAnts-p 


11 








— = 


: =: =: 


:; z^ := =r r= ;r =: 


^ '^-^A'^TZ 


f^ 


--^ 





The Number of 


■ Experts Assignm 


g tne opt^u 




Relative 
Ranking 


Figure 
20 


Figure 
21 


Figure 
22 


Design 


1 
2 
3 


10 
7 



8 
5 
4 




1 

16 


Operations 


1 
2 
3 


3 
2 



2 
3 






5 


Research 


1 
2 
3 


2 
3 



3 
2 





5 



D-20 



4. In conjunction with exits requiring two lanes, the number of lanes 

on the mainline beyond the ramp should be reduced by one. (Always 
or Never?) 

Always Usually Occasionally Never 

Design 2 10 4 1 

Operations 4 1 

Research £ _i 1 

Total 2 16 7 1 



5. The mainline traveled way should never be reduced by more than one 
traffic lane at a time^ (Always or Never?) 

Always Usually Occasionally Never 

Design 7 8 1 

Operations 2 3 

Research _A ^ _2. __^ 

Total 13 12 1 



6. When a lane is to be dropped in the vicinity of a major inter- 
change, it should be carried beyond the interchange and then 
terminated beyond the influence of the interchange » (Always or 
Rarely?) 

Always Usually Occasionally Never 



Design 


2 


11 


3 


1 


Operations 


2 


1 


2 





Research 


2 





2 






Total 6 12 



7. When a lane is dropped beyond an interchange, in general, it 

should be carried beyond the end of the entrance ramp accelera- 
tion lane for a distance of: 

3,000 
0-500 fto 500-1,000 ft. 1,000-2,000 ft, 2,000-3,000 fto ft, + 

Design 2 7 6 

Operations 3 2 

Research 1 1 JL Ik 

Total 1 3 11 6 3 



D-21 



8. Ideally, lane drops should be located at major diverging forks, 
(Agree or Disagree?) 



Design 

Operations 

Research 

Total 



Strongly 
Agree 

2 





Agree 

9 
3 

_2 

14 



Disagree 

4 


1 



Strongly 
Disagree 

1 

2 



Assuming that for appropriate reasons a lane must be dropped 
beyond an interchange, please rank these four configurations 
in the order that you personally prefer to employ them. 



^- — 4 Lanes 




3 Lanes —-j 


-^ 






^Interchange -♦ 






"*■ 1 


28 




'^^ Drop Right Lane 


T^^ — 4 Lanes 




1 Drop Left „ ^ 

1 ^ 3 Lanes 7 

/ Lane / 


^Interchange " -♦ 






-^ 






-^ 


29 


\— 4 Lanes 




3 Lanes — j 






■ 


ir — V 1 


-^ 


N 




30 




— Merge Middle Lanes 


\- — 4 Lanes 


^ 


> Drop Lane 3 

y^ 3 Lanes — -7 


•^interchange -«- • — =^ — . 1 


-^ 




1 


-». 1 


31 







D-22 



The Numbez of Experts Assigning the Specified Rank 







Figure 


Figure 


Figure 


Figure 




Rank 


28 


29 


30 


31 




1 


11 











Design 


2 





10 


1 







3 








9 


2 




4 





1 


1 


9 




1 


3 











Operation 


2 





3 










3 








3 







4 











3 




1 


4 











Research 


2 





3 










3 





1 


3 


2 




4 








1 


2 



Note: These responses were given by the participants of jt:he second 
workshop only. 



D-23 



10. Assuming that at a major fork two lanes are to be dropped, 
please rank these two configurations in the order that you 
personally prefer to employ them. 



-x 



4 Lanes 



2 Lanes 



7 




V 



4 Lanes 



(Drop Lane Beyond Interchange) 
3 Lanes — -^ 




33 



Optional Lane 



The Number of Experts Assigning the Specified Rank 



Design 

Operations 

Research 



Rank 

1 
2 



Figure 
32 

6 
4 

2 

1 

2 
2 



Figure 
33 

4 
6 

1 
2 

2 
2 



D-24 



11. In general, what taper rates do you feel are most appropriate for 
lane drop treatments beyond the interchange (assuming that there 
are valid reasons for dropping the lane beyond the interchange)? 



60 MPH 


SPEED 
















Minimum Rate _?_: 


1 




Desirable Rate 


1* 1 




25 35 40 


50 60 


70 


40 


50 


70 75 80 


90 100 


Design 

Operations 

Research 


13 
1 
10 


4 1 

2 


2 






1 


4 

1 
2 


2 12 




1 1 





Total 



70 MPH 


SPEED 














Minimum 


Rate ?_: 


1 




Desirable Rate 


1* 1 




35 50 


70 80 




50 


55 60 70 75 


80 100 


Design 

Operations 

Research 


1 5 

1 

1 2 


3 1 






2 

2 


12 
1 
10 


1 4 





Total 



Note: These responses were given by the participants of 
the second workshop only. 



12. If overriding width restrictions make it necessary to drop a lane 
just past an exit terminal, what design configuration do you feel 
should be used? 

a. A taper beginning at the exit gore nose (specify taper rate, 

b. A full width recovery lane followed by a taper (specify 
length and rate) 

c. There are no restrictions or conditions which justify dropping 
a lane just past an exit terminal. 

a. 



Design 
Operations 
Research 
Subtotal 

Total 





Taper 


Rate 


(?:1) 





35 50 


55 


70 100 


1 




1 


1 3 


1 

1 4 




1 
1 


1 1 




1 1 



D-25 



Length (ft.) Taper Rate 

150 360 650 800 1,000 25 30 50 70 

Design 1001 2 0031 

Operations 0000 1 0000 

Research 0110 1110 

Subtotal 1111 3 1141 



Total 



Note: These responses were given by the participants of 
the second workshop only. 



I 



D-26 



ROUTE CONTINUITY 



Workshop //I ONLY 

1. Which variable should control in the decision to make one movement 
an exiting movement and the other a through movement? 

a. The numbered route should always be designed as the through 
road regardless of turning volume. 

b. The relative voliimes of movements determine which is through 
and which exits. If the route changing volume is substantially 
greater than the continuing route traffic, the changing route 
volume should be designated the through movement. 

a. b. Total 

Design 3 6 9 
Operations 112 
Research 0^ 1^ __! 

Total 4 8 12 



2. If you checked (b) of question 1, please indicate below the rela- 
tive volume which must occur before the (b) alternative is used. 



_ ^. Continuing Route Volume 

Ratio =» -rr r- — *— : — vVi ' " 

Changing Route Volume 



a. ^1.00 > Ratio >_ .90 

b. .90 > Ratio >^ .80 

c. .80 > Ratio >_ ,70 

d. .70 > Ratio > .60 

e. .60 > Ratio >_ ,50 

f. .50 > Ratio >_ .40 

g. .40 > Ratio >_ .30 

h. ,30 > Ratio >_ .20 

i, .20 > Ratio >_ ,10 

j. .10 > Ratio 



d. e, f. g. h. i. j. Total 



Design 0001022100 6 

Operations 0000010000 1 

Research 0000000100 1 

Total 0001032200 8 



D-27 



3. Should all weaving be confined to collector-distributor roads on 
major interchanges? 

Yes No Total 



Design 


7 


2 


9 


Operations 


2 





2 


Research 












Total 9 2 11 



Workshop //2 ONLY 

1. Which variable should control in the decision to make one movement 
an exiting movement and the other a through movement in an urban 
area with high volumes? 

a. The numbered route should always be designed as the through 
road regardless of turning volume, 

b. The relative volumes of movements determine which is through 
and which exits. If the route changing volume is substantially 
greater than the continuing route traffic, the changing route 
volume should be designated the through movement. 

a. b. Total 

Design 2 7 9 

Operations 2 1 3 
Research 3 2 5 

Total 7 10 17 



2. If you checked (b) of question 1, please indicate below the relative 
volume which must occur before the (b) alternative is used. Per- 
centages refer to total volume approaching a split. (Note that it 
is still on urban area.) 

a. 90% continue, 10% change rpute 

b. 80% continue, 20% change route 

c. 70% continue, 30% change route 

d. 60% continue, 40% change route 

e. 50% continue, 50% change route 

f. 40% continue, 60% change route 

g. 30% continue, 70% change route 
h. 20% continue, 80% change route 
i. 10% continue, 90% change route 

a. b. c. d. e. f. g. h. i. Total 

Design 003000110 5 

Operations 000000100 1 

Research 19.9.11kkll 1 

Total 003001310 8 



D-28 



3. It appears that for rural major interchanges where voltimes are not 
high it is preferable to take the continuing-on-the-same-route 
traffic through and the changing route traffic off on a connection. 
Do you agree? 

Yes No Total 



Design 


8 





8 


Operations 


3 





3 


Research 


5 





5 



Total 16 16 



4. At what total approach volume would you consider discarding the 
route continuity approach and consider letting the volume splits 
determine the through road? 

ADT Approaching 



ADT Number Responding 

15,000 1 

30,000 1 

20,000 - 25,000 1 

Total 3 



5. If a weaving section within a major interchange is unavoidable 
because two adjacent loop ramps had to be used in the design, 
should a collector-distributor road be used? 

Yes No Total 



Design 


8 


1 


9 


Operations 


2 


1 


3 


Research 


5 





5 



Total 15 2 17 



6. In urban areas where ramps are often closely spaced is it often 
difficult to get adequate weaving distances between the major 
interchange and the nearest upstream or downstream ramp? 

Yes No Total 



Design 


9 





9 


Operations 


5 





5 


Research 


2 





2 



Total 16 16 



D-29 



7. In your experience has the problem of weaving between interchanges 
as described above ever resulted in design changes which produced 
undesirable operations at a major interchange? 

Yes No Total 

Design 4 3 7 
Operations 10 1 
Research 2_ 0^ v __2_ 

Total 7 3 10 



8. In a rural area with practically no land restrictions and very low 
turning movements would a design utilizing loop ramps which would 
produce a weaving section on a collector-distributor road be a 
feasible alternative? 

a. Yes 

b. Probably not, but under special circumstances I would consider it 

c. I would never consider loops and weaving sections in a major I/C 

a. b. c. Total 

Design 8 10 9 

Operations 111 3 
Research _4 £ 1_ _5^ 

Total 13 2 2 17 



9. If you circled (b) above please describe the circumstances which 
would have affected you. 

Comments : 

1, Weaving on a tight vertical and/or horizontal alignment. 

2. Sometimes, even in a rural area, existing routes requir- 
ing an interchange are spaced closely. Such proximity 
might cause you to place both interchanges on a C-D road. 
This case is rare with freeway-to-freeway, but not so with 
local access interchanges. 

The kind of traffic; e.g., commuter and adjacent designs. 



D-30 



I 



APPENDIX E 



DECISION THEORY APPROACH TO INTERCHANGE DESIGN 



John C. Hayward, The Pennsylvania State University 
James I. Taylor, The Pennsylvania State University 

E-i 



CONTENTS 

Page 

Introduction to Bayesian Decision Theory E-1 

Previous Applications E-17 

Interchange Evaluation in Decision Theory 

Framework E-21 

A Decision Theory Method for Interchange 

Evaluation E-25 

Example Problem E-37 

Implementation Difficulties E-52 

Trade-off Analyses Extension E-54 

Conclusions . E-75 



E-ii 



FIGURES 

Page 

E-1 Example Payoff Matrix E-4 

E-2 Filled Payoff Matrix E-5 

E-3 Sample Utility Function for Money E-10 

E-4 Example Probability Density Function E-14 

E-5 Example Cumulative Distribution Function E-14 

E-6 Payoff Matrix for Interchange Problem E-24 

E-7 Example Goal Structure and Performance Measures . . . E-27 

E-8 Sample Weighting Procedure . E-32 

E-9 Sample Payoff Distributions E-36 

E-10 Sample Payoff Distributions with Weighting Bands . . E-38 

E-11 Performance and Worth Distributions E-40 

E-12 Example Total Worth Distributions E-51 

E-13 Alternative Exit Ramp Configurations E-64 



E-iii 



TABLES 

Page 

E-1 Sample Weighting Schemes E-33 

E-2 Example Problem Performance Measures 

and Weighting Schemes E-39 

E-3 Example Problem Sample Values for 

One Sample E-48 

E-4 Example Problem Total Work Rankings E-50 

E-5 Results from Questionnaire on Trade-off 

Analysis » . . . . E-5 7 

E-6 Median Merit Ratings for Exit Ramp 

Configurations , E-66 

E-7 Median Meirt Ratings for Acceleration Lane Lengths . . . E-68 

E-8 Results from Questionnaire on Level -of -Merit Design 

Concept E-70 



E-iv 



APPENDIX E: DECISION THEORY APPROACH TO INTERCHANGE DESIGN 
Introduction to Bayesian Decision Theory 

One criticism of the Interchange design process Is the apparent 
lack of an acceptable evaluation method for choosing among alternative 
designs. This void in technique is disconcerting for two reasons. 
First, the possibility exists that a wrong selection will be made and 
interchange which does not operate properly, or which has a disastrous 
effect on the area around it, will be constructed n The second reason 
for concern is that the highway designer has no effective means for 
communicating the logic behind his decision to both his superiors and 
to the public. 

This appendix describes one approach for evaluating design alter- 
natives which should reduce the possibility of a wrong decision and 
enable the decision maker to better explain his reasoning for the 
final choice. The approach draws heavily on Bayesian management deci- 
sion theory developed largely at the Harvard Business School and 
recently applied to business decisions by many large corporations. 

Decisions Under Uncertainty 

There are two basic types of decisions; decisions based on complete, 
accurate information, and decisions based on uncertain information. 
It is in the latter category that evaluations and, ultimately, choices 
between alternative interchange designs must be made since so many of 
the projected effects of the facility are unknown or can only be grossly 
predicted. The designer makes his decision on the basis of incomplete 
and largely uncertain information. 



E-1 



The advocates of the decision theory approach recognize that the 
choice has to be made under uncertainty and seek to structure the 
problem so as to incorporate estimates of the uncertain factors 
rather than ignoring them. To ignore uncertainty would mean that the 
decision would be based only on calculatable effects, not on immeasur- 
able ones. A decision based purely on a benefit/cost ratio calculated 
using only user time savings and construction and maintenance costs is 
one example of ignoring immeasurable (but no less real) factors. The 
decision theory approach seeks to identify all factors which have rele- 
vance to the decision and to explicitly judge what effect each factor 
will have on each alternative. The assumption is that by explicitly 
stating factors and values, the decision maker can approach the problem 
in a more systematic fashion, leading to a better understanding of the 
problem and, therefore, increased confidence that the resulting deci- 
sion is correct. A decision theory approach seeks not to replace judg- 
ment, only refine it . 

Two Concepts of Probability 

Probability can be thought of in two contexts, mathematical and 
subjective. The mathematical concept of probability relies on frequency 
data to produce expected percentages of occurrence. This classical sta- 
tistical view of probability can be illustrated by the example of red 
and green balls contained in an urn. If we know that there are 10 red 
balls and 20 green ones, we consider that the "probability" of randomly 
drawing a red ojie is 1/3. This mathematical probability can be computed 
when the quantities or frequencies of different events are known. 

The Bayesian statistician admits to another kind of probability 
labeled "subjective probability." This kind of probability is derived 

E-2 



from a person's intuition about a particular event which is about to 
occur but whose outcome cannot be predicted mathematically. An example 
of this type of probability can be illustrated by asking a person what 
he feels the chances are that it is going to rain tomorrow. If he says 
that he feels that there is a 35% chance of rain, then he has given a 
subjective probability of .35 that it will rain. A Bayesian statistical 
approach will admit this kind of subjective information into a subse- 
quent analysis or will, in effect, place a value on the decision-maker's 
judgment . 

Obviously, there are some problems which are better analyzed using 
mathematical uncertainty, other problems which require judgmental proba- 
bilities and still others which need both kinds of analysis. A labora- 
tory experiment where all factors can be adequately controlled can best 
be analyzed by measuring effects and then making inferences based 
strictly upon one's observations. In the uncontrollable "laboratory" 
of the highway system, we must often resort to subjective approaches 
as the commonly used "diagnostic study team" evaluation method illustrates, 
The point is that until mathematical probabilities can be accurately mea- 
sured and replicated, the decision maker must make use of subjective pro- 
babilities or, as more commonly referred to by designers, engineering 
judgment. 

Payoff Matrices 

The basic tool of a decision theory analysis is known as a payoff 

matrix or payoff table. The matrix is two dimensional, with one side 

being described by the set of alternative choices available and the 

remaining side being described by various "state of nature" or uncertain 

future events, only one of which will occur. 

E-3 



A simple example is the rain example again. The set of alterna- 
tive actions could be: (1) to carry an umbrella, or (2) not to carry 
an umbrella. Two states of nature may be: (1) it rains, or (2) 
it does not rain. The payoff matrix (without payoff entries) is 
shown below as Figure E-1. 

ALTERNATIVE ACTIONS 



(1) Carry 
Umbrella 



(2) Do Not Carry 
Umbrella 



STATES 
OF 

NATURE 



(1) Rain 



(2) No Rain 



Figure E-1. Example Payoff Matrix 



One necessary stipulation in formulating a decision in this 
manner is that the states of nature must be independent of the alter- 
native actions or, in our case, that carrying an umbrella will not 
cause it to rain. 

The cells of the matrix are filled in with payoffs of each com- 
bination of action and state of nature. The unit of payoff most readily 
brought to mind is the dollar, and although not always the most appro- 
priate, we will use it here for simplicity. Suppose that if we carry 



E-4 



the umbrella and it rains that we will not be paid anything (but 
we don't have to pay out anything either), so we value the "payoff" 
at $0. On the other hand, if we don't carry the umbrella and it 
rains, our suit may be ruined at a cost of $100 or a payoff of -$100, 
Carrying an umbrella when there is no rain has some cost associated 
with the inconvenience — assumed here to be $2.50; That is, the 
payoff is -$2.50. Finally, if we choose not to carry an umbrella 
and it doesn't rain our cost and our gain are zero. The filled 
payoff matrix is shown in Figure E-2. 

ALTERNATIVE ACTIONS 



STATES 

OF 
NATURE 



(1) Rain 



(2) No Rain 



(1) Carry (2) Do Not Carry 
Umbrella Umbrella 



$0 


-$100.00 


-$2.50 


$0 



Figure E-2. Filled Payoff Matrix 



The final inputs required to complete this simple example are 
called prior probabilities and are used to represent the uncertainty 



The term "prior" is consistent with the literature but its meaning 
may not be clear to the reader. These "prior probabilities" are 
often modified through experimental results before the analysis is 
completed, hence the qualifier, prior. 



E-5 



regarding the states of nature „ These are subjective probabilities and 
may be obtained from expert judgment (the weather forecast) or from a 
novice impression (looking at the sky in the morning) . Assume that 
in the example, the probability of rain is estimated to be 0»20 and 
that the probability of no rain is 0.80. The restriction on setting 
prior probabilities is that they must sum to unity, but this is usually 
handled by careful definition of the states of nature so as to 
include all possible states. In our example, one would not choose 
as states (1) rain and (2) sunshine, since there are many other possi- 
bilities. Rain and no rain cover all possibilities, however. 

The decision rule generally followed is to choose the alternative 
which maximizes expected payoff. One essentially computes an average 
payoff, weighted by the prior probabilities, for each alternative 
and selects the action which gives the highest expected value. 

The computation for our example is simple but illustrates the 
concept. Considering alternate (1), carrying the umbrella: 

Expected Payoff (carry umbrella) ~ Prob (Rain) x Payoff (carry 

umbrella, rain) 

4- Prcb (No Rain) x Payoff (carry 
umbrella, no rain) 

Expected Payoff (carry umbrella) = ,2($0) + .8 (-$2. 50) = -§2.00 

Expected Payoff (don't carry) = ,2 (-$100) + 8(0) = -$20.00 

Applying the decision rule which says to act so as to maximize payoff, 
we would choose to carry the umbrella — thereby selecting the $2,00 
expected loss (-$2.00 expected payoff) over the $20.00 expected losso 

Describing the payoff matrix mathematically, let the set of n 
alternative actions be represented by a, , and the set of m states of 

E-6 



nature be represented by 9 . Prior probabilities are given as P(6 ) 

■J J 

and payoffs as R(a., 6 ). The payoff matrix reduces to 



m 



R(aj^, 9^^) R(a2, 9^) 



R(a^, 62) 



^^^r V 



• « • 



n 



R(a^, 9^) 



R(a , 9 ) 

n m 



The expected payoff for each alternative is computed by the formula; 



m 



ER(a^) 



E P(9 ) R(a, 9J. 
j=l -■ ^ J 



where ER(a.) » expected payoff for alternate i. The decision rule to 
maximize expected payoff would be: 



Choose a so that ER(a ) = Maximum [ER(a,), ER(a^) , , . . ER(a )1 



Utility as a Payoff 

In the simple example of the decision whether to carry an umbrella 
or not, the unit of payoff was assumed to be money. In most complex 
decisions dollar costs and benefits are not the most appropriate units 
to use in the payoff matrices for two reasons. First, many evaluative 
categories or attributes of a particular alternative state-of-nature 
cell cannot adequately be expressed in dollar terms. An example might 
be the choice of sending one's son or daughter to one of three different 
colleges. Each cell has an entire set of payoff attributes, all of 



E-7 



which must be considered in the evaluation. Some are quantifiable in 
dollar terms, such as the costs of tuition and room and board, but 
many attributes such as quality of instruction, exposure to undesirable 
elements, or stress on athletic programs, are not amenable to expression 
in relative dollar amounts. This means that the final payoff must be 
expressed in something other than money and the decision criteria must 
seek to maximize this other unit. 

The second reason why money is a poor indicator of the value or 
worth of an alternative choice is that the value one places on money 
is not linear. This may be illustrated by considering a betting situa- 
tion where one chooses to bet on a football game. Suppose you are 
given the opportunity to bet one dollar, to bet $1,000, or not to bet 
at all. If you think that team A will win with a probability of ,60^ 
then your expected dollar payoffs will be: 

Action Expected Payoff 
Bet $1 $ .20 

Bet $1,000 200.00 

Don't Bet 0.00 

If you act to maximize expected dollar payoffs you will choose to 

2 
bet $1,000 on team A to get a $200 expected payoff. 

Many people would choose not to take the $1,000 bet unless they 

were relatively unconcerned with the 40% chance of losing $1,000. 

Therefore, the assumption is that they must be acting not to maximize 

expected dollars, but rather to maximize the utility or worth of 

money . 



^The payoff is equal to .6($1,000) + .4(-$l,000) = $200. 



E-{ 



Utility can be defined as a unitless measure of the relative worth 
or value of different actions under a given state of nature. If a per- 
son finds an orange more appealing than an apple, the utility of an 
orange is greater to him than that of an apple. 

In general, quantities of each payoff affect the utility value 
given to the attribute in a nonlinear fashion. The example of money 
can be given again. 

Suppose one were to measure a person's utility for incremental 
amounts of money from zero to a million dollars. The graph of utility 
versus money, called a utility function for money, may look like 
PigVire E-3. The endpoints are simple to fix; a person has the most 
utility or assigns greatest worth (1.0) to obtaining $1 million and 
least (0.0) to obtaining zero dollars. The curve goes up more rapidly 
at first than it does later on. There is more difference in utility 
between gaining half a million and gaining zero, than there is between 
half a million and a million. This general shape of curve represents 
the economist's notion of diminishing marginal utility for money, 
although it may vary in the deceleration constant. It will probably 
be flatter for millibnai*;©* » 

This notion of utility or worth of some measured quantity gives 
the decision maker a normalizing scale for comparing unlike quantities. 
For an action which has more than one kind of payoff (e.g., safety 
improvement which will affect accident rate and severity) it allows 
one to reduce the payoffs to a common scale. Utility alone does not 
give trade-off information about the worth of a fatality versus the worth 
of an injury accident, but only the worth of a number of fatalities ver- 
sus another number of fatalities. Trade-off can be incorporated into the 

evaluative methodology later. 

E-9 



1.0 -- 




$500,000 
MONEY 



$1,000,000 



Figure E-3. Sample Utility Function for Money 

Utility also gives the decision maker a tool for assessing or 
ranking those immeasurable categories which must often be considered 
in the analysis. That is, those categories which have no measuring 
scale, such as neighborhood disruption or visual impact on the non- 
user, can be directly estimated as point values on a utility scale. 
In effect, this amounts to a rating of certain variables for each 
alternative on a utility rating scale. 

This concept gives one rational approach to the twofold diffi- 
culty of incorporating the two types of immeasurable evaluative cate- 
gories discussed in Chapter Two. First, it coverts all evaluation 
parameters onto a common scale so we may compare dollars to decibels, 
for example, and second, it allows us to directly assess the value of 
those variables without universal scales of reference. Ultimately, 
we arrive at a measure common to dollars, decibels, and visual impact, 
and any other evaluative category one may think of. 



E-10 



I 



The utility concept further implies that one makes his decision 
using a criterion of maximizing utility rather than payoff units. This 
logic has been substantiated through experimental devices and can be 
supported intuitively by considering various kinds of betting situations 
such as the football game example. 

Asseeain^ Pyioy Probability and Utility Functions 

Two major inputs to the decision theory framework described above 
are prior probabilities and utility functions. It is tempting to 
ignore discussion of the serious difficulties encountered in assessing 
probabilities and utility. In fact, many of the researchers in the 
field have preferred to dwell on the more intricate mathematical 
nuances of Bayesian statistics and have assumed that the inputs, pro- 
babilities and utilities, would come easily. This step in the analysis 
is peirhap? the most cruciall to obtaining good results, however, and 
should not be passed over lightly in the development of an implementable 
procedure. 

Prior Probability Assessment 

Several approaches can be taken in assessing an individual's prior 
probabilities, but the best one seems to depend upon the decision 
foaker's personal background. The analyst (the individual who is 
Hjrying either to obtain the values so that he may make the analysis him- 
self to teach the decision maker to do it) must work closely with 
the decision maker, explaining the concepts, leading discussions to 
determine prior probabilities, constructing a distribution, and then 
obtaining the decision maker's approval. The process is an iterative 
one, with the prior distribution being refined until the decision maker 

le satisfied that the distribution reflects hia best judgment. 

E-11 



F^ve assessment methods which may be used to obtain a decision 
maker's probabilities on states of nature are discussed briefly on 
th^ following pages. 

(1) Direct assignment 

This method is the easiest for those decision makers who are exper- 
ienced in the use of decisiqn theory or who have enough statistical 
background to fully understand probabilities. If the states of nature 
are discrete and defined fully, the analyst simply asks for the pro- 
bability of egch occurring. 

(2) Betting odds 

Other decision makers may feel more comfortable expressing their 
prior probabilities as betting odds, particularly when there are only 
two possible states of nature (A and B) . If the decision maker indi- 
cates that the odds are 3 to 1 that A will happen instead of B, he is 
assigning a 0.75 probability to A and a 0.25 probability to B. 

(3) Lottery methods 

This approach involves asking the decision maker to choose between 
tW9 lotteries which are constructed so as to assess his prior proba- 
bility. For example: 

Lottery A. You win $25 with probability 0.35, 

or you win $0 with probability 0.65. 
Lottery B. You win $25 if state of nature 6 occurs, 

or you win $0 if state of nature 9. does not 
occur. 
If the decision maker (DM) chooses Lottery B, he must feel that 
the probability of state of nature 6^ occurring is greater than 0.35. 
If the analys't then varies the probability of winning in Lottery A until 

E-12 



the decision maker is indifferent between A and B, the final pro- 
bability of winning $25 in Lottery A is a measure to the DM's prior 
probability for state of nature 9^ . 

(4) Probability density function 

Many times states of nature are continuous rather than discrete. 
For example, consider thdt a set of "states of nature" is a traffic 
forecast and the decision deals with how much capacity to provide. 
The prior probability on the states of nature (individual valvies pf 
traffic forecasts) can be expressed as a continuous probability density 
function similar to the one in Figure E-4. Decision makers may be able 
to draw the density function with help from the analyst if they aire 
comfortable expressing uncertainty as probability distributions. 

The probability of the actual value falling between two points 
is shown as the area under the curve between those two points on 
the abscissa. Therefore, in Figure E-4 the probability that the 
actual traffic T will be less than or equal to 14,000 can be repre- 



sented by: 



or, in general: 



14,000 

P(T < 14,000) = / f(t) dt 
^ 10,000 



X 

P(T < X) = / f(t) dt. 
a — 

o 



The restriction on f (t) is that the area under it sums to unity, 

00 

/ f(t) dt = 1 



O 



E-13 




10,000 



14,000 15,000 
TRAFFIC FORECAST, t,( A DT) 



20P00 



Figure E-4. Example Probability Density Function 



Q. 



< 

CD 
O 
(T 
0. 




10,000 



15,000 
TRAFFIC FORECAST, t.(ADT) 



20,000 



Figure E-5. Example Cumulative Distribution Function 
E-14 



(5) Cumulative distribution function 

Another way to represent continuous probability is with the cumula- 
tive density function, or a function which represents the change in 
area under a probability density function as the variable of interest 
is increased. The traffic forecast prior probability is shown in 
Figure E-5 as a cumulative function. This gives the probability that 
the actual value, T ,will be less than some value of t directly off the 
vertical axis. 

Some decision makers can relate to this kind of continuous repre- 
pentation better than the probability density form, although both repre- 
sent the same thing. The analyst must elicit the information from the 
DM by asking him how often he feels the actual value will fall below 
some specific value within the range of the distribution. The answer 
is an indication of the prior probability of something less than that 
value occurring. 

Utility Function Assessment 

First, it must be noted that we are generally interested in a 
utility function for an attribute measure bounded by some prescribed 
limits. Suppose we consider the utility of money, as represented by 
a function such as depicted in Figure E-3. How could one determine such 
a function? 

To assess utility functions, the analyst most commonly employs a 
lottery scenario with the decision maker. For many attributes, like 
money, it is obvious which end of the scale is preferable over all 
other points and also which is least preferable. These form the 
starting points for the utility function construction by serving as the 
maximum and minimum values on the utility scale. In the case of money 

E-15 



between zero and $1 million where the utility scale is from to 1, the 
zero amount is given the zero and the highest incremental amount the 1. 
For the remaining points in the money example, the decision maker 
is offered his choice of two lotteries. He may take x dollars for cer- 
tain (Lottery A), or enter a lottery where he can win $1 million with 
probability P or win zero with probability (1-P) , The P value where 
he is indifferent is the utility value for x dollars. In the actual 
interviewing process either x or P is varied with the other held con- 
stant until the DM says that both lotteries are equally appealing to 
him, 

The logical proof of this method rests in the assumption that people 
behave so as to maximize expected utility. If the DM says he is indif- 
ferent between Lotteries A and B, he is indicating that the expected 
utilities of each are equal. Consider the following example of after 
tax gains: If a subject is indifferent between getting $500,000 with 
certainty and winning $1,000,000 with a probability of .8 (and winning 
zero with a probability of .2), his utility for $500,000 is .8 (relative 
to utility of $1,000,000 set at 1.0). In a proof form consider the 
two lotteries: 

Lottery A; Get $500,000 for certain 

Lottery B: Win $1,000,000 with probability .8 or win 

zero with probability .2 
If indifferent: Expected Utility of A = Expected Utility 

of B 
Expected Utility of B = .8 (Utility for $1,000,000) + .2 

(Utility for zero) 
= .8(1) + .2(0) = .8 

.'.Expected Utility A = 0.8 = Utility for $500,000. 

E-16 



Several points on the curve can be determined in this fashion and 
the function drawn in. 

For the utility functions which describe discrete and unlike objects 
the DM is first asked to rank the objects. The most preferred is given 
a utility of one and the least preferred is given a value of zero. 

An example might be the utility function one has for several types 
of comparable automobiles; Chevroletj Ford, Plymouth, Chrysler, or 
MfSrcury. He ranks them as to preference in the following manner. 

Most Preferred (1) Mercury 

(2) Chrysler 

(3) Plymouth 
(4J Chevrolet 

Least Preierred (5) Ford 

The Mercury is a.'^^igned a one and the Ford a zero value of utility. He 
then asks himself where he is indifferent between a Chrysler for cer- 
tain or a P chan.:e on a Mercury with a (1-P) chance on a Ford. The 
value of P repxe.'-^ent^ his u^^ility for the Chrysler car. 

Both the money exc;.mpie and the car example are simple ones to 
demonstrate how ur.iiit}' functions are arrived at. The method assumes 
that one can rank h\s preferences, ac least as far as selecting one as 
the best and another as the poorest ^ The second premise is that a 
decision maker chooses to maximize his expected utility, not the 
expected value of the two lotteries. 

Previous Applications 
Will a decision theory approach improve the quality of complex deci- 
sions and is the imprcvemenc worth the trouble? It would be ideal to answer 

E-17 



such a question with a review of past successful applications of the 
methods outlined above, which would prove that wrong decision had 
been avoided and costly errors eliminated. Unfortunately, such 
studies are really impossible to perform since most complex deci- 
sions are one of a kind and after they are made and implemented 
one doesn't know for certain that the best choice was made. Every 
major interchange involves a unique decision or choice and after a 
configuration is chosen and built the designer-decision maker does not 
have the opportunity to try another design. Therefore, it is impos- 
sible to design an experiment which proves the value of the decision 
theory approach. 

The measure of the value of a decision theory approach must come 
from the users of the methodology, not from comparisons of choices 
with and without the method. The justification for adoption can be 
made by outlining where it has been tried and what the people who 
tried it thought of it. Applications have been made primarily in 
the business fields, in military decisions, and the field of 
medicine. Methodologies have been proposed for use in transport 
planning decisions and a few for interchange design. The business 
applications xd.ll be discussed below, followed by a brief section 
on proposed transportation planning methodologies. 

Business Applications 

An interesting review of decision theory applications in business, 
written by R. V. Brown,, appeared in the Harvard Business Review in 



E-18 



1970 (Brown, 1970). The author surveyed 20 companies in 1969 to deter- 
mine what Impact decision theory analysis has had on decision and 
what difficulties have been encountered in application. All of the 
companies were known to have used decision theory approaches. 

He found that use of the technique is expanding rapidly due to 
initial "successful" application by a few pioneering companies (DuPont, 
Pillsbury, and General Electric) as well as increased production of 
decision theory-trained business school graduates, principally from the 
Harvard Business School. Brown discovered that adoption of the tool 
has caused little change in the decision-making process, but it has 
affected individual decisions. 

Some problems are encountered in the application of decision theory 
techniques to business decisions. It is not applicable to all pro- 
blems and the user must expect some disappointing experience at first. 
The company must have competent practitioners who can deal effectively 
with the executives that are doubtful of the value of the method. 
Often the logic and language of the procedure is new and uncomfortable 
for the decision maker. If the technique is pushed by the "front 
office," the lower level decision makers feel threatened and resist 
using it. If the analysis is performed by a staff group without 
much personal contact, the decision maker may feel that his power 
is being usurped and that he no longer has control over the decision. 
The cost of the procedure is a real consideration since many of the more 
sophisticated analyses require computer applications. 

In spite of the problems, the trend in business analysis seems 
to be moving toward a more widespread use of the technique. This 
trend would lead one to the conclusion that the decision theory 
approach is sound and leads to good decisions. 



E-19 



Transportation Planning Methodologies 

Several schemes using decision theory concepts have been proposed 
for use in transport planning evaluation. Though the majority of the 

work has been aimed at evaluation of alternative systems of a more 

3 
macroscopic scale than interchange desigp, a few have been directed 

A 
specifically at evaluating alternative interchanges. All of the metho- 
dologies follow basically the same logic. 

The first step is to determine those attributes upon which the 
alternatives will be evaluated. This can be done either through a formal 
or informal hierarchical goal development process where the "super goal" 
of improving the quality of life is broken into increasingly finer sub- 
goals until a list is arrived at which can be adequately represented 
by performance measures^ 

After the list of attributes has been agreed upon, performance 
measures or goal quantifiers must be established. For the goal of 
minimizing fatal accidents, for example, the performance measure might 
be the expected number of fatal accidents reduced from the "do-nothing" 
case for a particu.ar alternanive action. 

Some attribucBS have no scale which adequately measures or quanti- 
fies the degiee to wh3-ch the goals are met. For these, an artificial 
scale is constructed which can be used to rank the alternatives. 

Performance measures other than direct worth are transformed into 
utility or worth via some utility function, thereby reducing all evalua- 
tive attributes to the same scale. This enables the analyst to add the 



3 

See (Miller, April 1969), (Institute for Analysis, Sept. 1971), and 

(Manhelm, 1970), 

4 
See (Alexander and Manhelm, 1965) and (Leisck May 1972) . 

E-20 



utility score based on some weighting scheme which is intended to express 
the relative importance of each attribute to the total. In some cases 
benefits and costs are considered separately and the resulting over-all 
worth measure is a ratio. In other examples the worth values of 
the costs are simply added to the worth of the benefits and the total 
worth becomes a summation of the weighted values. 

Interchange Evaluation in Decision Theory Framework 

Having introduced the decision theory concepts of evaluation and 
pointed out some previous applications of the technique to business and 
transportation planning problems, it is desirable to define interchange 
evaluation as a decision theory problem. Each of the three components, 
alternative actions, payoffs, and states of nature, will be discussed 
individually- 

Alternative Actions 

It is obvious that the actions in the decision theory framework 
correspond to altemai:ive interchange designs. At one decision level 
one may consider an entire interchange configuration as an alternative, 
but at a finer level alternate actions might be restricted to different 
designs for one approach leg of an interchange. The actions available 
in the latter case may be as specific as the alternatives of placing 
an exit ramp on the left or on the right side of the roadway, while con- 
figuration actions may compare a cloverleaf to a direct connection con- 
figuration. The number of actions which may be considered are bounded 
only by the costs involved in working up the design to a level where 
decent predictions of performance measures may be made. 



E-21 



Payoffs 

The "payoff" of a particular interchange is a multi-dimensional 
one requiring a large set of units to describe adequately. P. P.M. 20-8 
requires that the interchange be evaluated over 23 separate categories 
pr attributes. Some of these have acceptable performance measures 
while others are qualitative attributes at best. Therefore, the inter- 
change payoff in a decision theory framework is a sgt of unrelated, 
individual performance measure estimates. 

This set approach or multi-attributed problem requires some nor- 
malizing approach which will reduce all evaluative categories to the 
same scale. At this point the utility concept comes ihtp play, p^ovid- 
^ng the needed tool for combining multiple attributes into a single 
unit system. 

The utility concept only provides an indication of the worth of a 
particular quantity of some attribute in terms of more or less of 
that same attribute, however. To get the total utility or worth, (the 
single value), a weighting scheme is required to obtain a weighted sum- 
mation of the individual attribute utilities. 

The payoff value for the interchange problem is derived via a four- 
stage process. First, the attributes and their performance measures 
are fixed and are held constant for all alternatives. Second, estimates 
of each performance measure for each alternative action are given in 
the appropriate units. Third, the performance measure units are trans- 
formed to utility or worth values via a utility function unique to that 
attribute. Finally, the utility values^ for each performance measure are 
summed according to a predetermined weighting scheme which allows between- 
attribute comparisons. The result is a single worth value (or, as yiH 

E-22 



be shown later, a distribution of worth values) for each cell of the 
payoff matrix. 

States of Nature 

The states of nature in an interchange problem may not be as easily 
visualized as the alternative actions or the payoff vglues, Xl>e metho-r 
dology presented below will consider different weighting schemes as thia 
states of nature. By different weighting schemes it is meant that 
there are many different ways to combine the many perfprmance-utility 
values in the payoff cell. That is, the feeling of relative importance 
between unlike attributes such as vehicular safety and non-user noise 
varies substantially from person to person. Ideally, we would like to 
use a weighting strategy which reflects the feelings of the entire 
society, but this kind of information is not currently available. 

An 5!,lternative to complete public assegsment is to devise several 
weighting schemes which may have support in different sections of the 
aociety. ^ach one of these schemes wpt4ld comprise a state of nature 
and could be represented by a prior probability. The prior probability 
WQuld be related co the percentage of society which feels that the parti- 
cular weighting scneme best reflects their Qwn opinion. For example, 
if the decision maker felt that 40% of the society ttiat was to be 
affected by the xncerchange favored a weighting scheme which was safety 
oriented then the prior probability of that state of nature would be 0.4. 
Another way of explaining a prior probability would be that the decision 
maker felt, with a 40% certainty, that the state of nature was the "true" 
feeling of the soceity. 



E-23 



Summary of Interchange Problem Framework 

Figure E-6 represents the interchange problem in the decision matrix 
form. 





9l 


Weighting 
Scheme 1 


STATES 
OF 


^2 


Weighting 
Scheme 2 


NATURE 


«3 


Weighting 
Scheme 3 




«4 


Weighting 
Scheme 4 



DESIGN ALTERNATIVES 
A B C D 



* 


'^BJ 


'^Cl 


'^D,! 


^A,2 


^B,2 


\,2 


•^0,2 


^A,3 


^^8,3 


^^0,3 


"^0,3 


^A,4 


S,4 


•^0,4 


•^0,4 



R. . is the cumulative worth of alternative i 
combined under weighting scheme j 



Figure E-6. Payoff Matrix for Interchange Problem 



If each of the "m" states of nature, 6,, has a prior probability of 
P(6.), then to maximize expected utility, the decision maker should 
choose the alternative, i, so that 



m 

E R.,. X p(e.) 
j=l ^ J J 



is the maximum value. 



E-24 



A Decision Theory Method for Interchange Evaluation 

The following section will outline a step-by-step approach to the 
interchange evaluation problem using the payoff matrix concept formu- 
lated above. 

Step 1: Establish a Goal Hierarchy 

The Important evaluative attributes for any decision can be logi- 
cally arrived at by considering the goals of the action as a hierarchy 
of increasingly explicit subgoals. The goal structure may be visualized 
as a tree which becomes more defined in its terminology as one moves 
out the branches. The lowest level subgoals become the attributes of 
the evaluation procedure and are later represented as performance 
measures. 

One goal hierarchy example in a transportation context is given by 
Manheim and Hall (.i968) . Their ultimate goal is called "the good life" 
which is characterized at the next lowest level by convenience, safety, 
aesthetics and economic considerations. Each of these four subgoals are 
broken down further; for example, safety means decreasing fatalities, 
decreasing injuries, and decreasing property damage accidents on the 
highway. By subdividing the super goal of the good life the authors 
derive twenty measurable subgoals or evaluative attributes. 

Structuring this type of hierarchy allows the decision maker to 

use different levels of goals for different decisions in the process. 

The level of detail upon which final design decisions should be based, 

for example, would be a much more explicit goal level than corridor 

choice decisions. The decision maker may consider "safety" in evaluating 

alternative corridors, while considering property damage, Injury and 

fatal accidents separately for a final design decision, 

E-25 



1 



The goal hierarchy exercise also eliminates using an attribute 
list which uses different levels of the same goal as separate attributes. 
For example, the three different types of accidents are subsets of the 
goal, "safety." All four, safety, fatal, injury, PDO accidents, should 
riot be considered simultaneously in the evaluation; rather the attri- 
bute should be either safety or the set of three types of accidents. 
(Level confusion is apparent in the list presented in PPM 20-8.) 

Step 2: Establish a Performance Measure for Each Lowest Level Go^i 

A performance measure must be adopted which reflects how closely 
each alternative design comes to satisfying the goal. For example, one 
goal may be to keep construction costs low, with the performance mea- 
sure in dollars. Another goal may be to keep the noise level in the 
community low, and the attendant performance measure might be decibels 
at some prescribed distance from the edge of pavement. 

It is desirable to express goals or goal attainment in terms of 
physical measures- Unfortunately, this is not always possible, either 
because no measr-jfe exists or the goal is not expressed at a fine enough 
level. An example of an attribute with no performance measure might 
be neighborhccd disruption. The goal of safety cannot be quantified 
directly without breaking it down further into different types of safety. 
In both these c<as5es the performance measure may have to be a direct 
worth estimate of the value of the alternative rather than a physical 
measure. By using direct worth estiamtes the need for transforming 
the physical measures into worth or utility measures is eliminated. 

Figure E-7 presents an example of the goal hierarchy concept 
and the matching performance measure notion. It is not intended to 
be a recomiriended format for all projects, but is given only to illustrate 

E-26 



1 



GOAL HIERARCHY (more explicit subgoals -»-) 



PERFORMANCE MEASURES 



"GOOD TRANS 
PORTATION" 



Convenient 
Transport 



.Safe 
Transport 



1 I 

-Low travel times Reduced time (min.) 

— Free flow Level of service (A-D) 

'-Reduce user stress Direct worth 



r-Low fatalities Reduced fatal accidents 

Low injuries Reduced injury accidents 

Low property damage Reduced PDO accidents 



r-User- 



.Aesthetical- 
ly Pleasing 



Low noise Decibels atfc 

Visually pleasing Direct worth 

•—Comfortable ride Roughness index (1-10) 



■-Non-User 



i 



Low noise Decibels at ROW line 

Visually pleasing Direct worth 

Low water pollution Percent increased (%) 

Low air pollution Percent increased (%) 



Beneficial 
— to the 
Conmunity 



Econom- 
ically 



4 



Increase industry Increased payroll (%) 

Decrease unemployment Increased jobs 

Increase tax base Increased assessed value 

Increase fire protection. . . . Decreased average time 



J— Improve neighborhood Direct worth 

<. . ,, H- Improve poor & aged mobility. . Direct worth 

■iociaiiy — 1 ji^p^Qyg recreation Direct worth 

•—Permit desired growth Direct worth 



_Low Cost 

Transport 



Construction Dollars 



Operating 



User cost Dollars /year reduced 

Facility cost Dollars/year reduced 

I I 



Fjigure E-7 . Example Goal Structure and Performance Measures 

E-27 



the output of a goal hierarchical structure and performance measure 
procedure. This example will be continued through the remainder of 
the Appendix E discussion. 

Step 3; Generate Alternatives 

Major interchange design is essentially a search and selection 
procedure where the generation of alternative designs constituting the 
search and the evaluation procedure is used to select the best of the 
alternatives. In such a process the concept of optimal design becomes 
meaningless since the design procedure will lead to an optimal facility 
only by chance. That is, the alternative set must include the optimal 
design before any evaluative technique can select it. Therefore, to 
Increase the probability of hitting upon the optimal design, the engineer 
may either increase his alternative set size or be more selective in 
his choice of alternative designs. Because increased >numbers of alter- 
natives mean added design costs and, according to the workshop experts, 
would increase the chances of optimal selection only slightly, the 
second approach i^ mere desirable. 

Alternative designs should be generated to portray the vide range 
of goals which appear in the goal structure. Each alternative design 
might be direcred primarily toward achieving one of the higher level 
goals' with seccndo-ry consideration of the remaining goals. For example, 
one alternative mighc be designed with safety as the ultimate goal, 
another stressing lew costs and still another may be intended to pro- 
vide ultimate user convenience. 

A second considferation in generating alternative solutions is 
that they sat:v = ty minimum performance measure standards. This feasibi- 
lity constraint insures that all alternatives which are to be evaluated 

E-28 



are In conformance with the policies and guidelines presented in the 
design manuals of the particular state. Such constraints might be 
in the form of a ceiling cost on construction or maintenance, minimum 
level of service requirements, or provision of adequate stopping sight 
distance. 

Step 4: Obtain Performance Distributions 

Given the set of alternatives and the measures and goals with 
which to evaluate them, the decision maker must predict how each 
alternative will "score" in each performance category. Often the 
decision maker, himself, is not in the best position to make predictions 
in all the evaluative categories, and he will assign the scoring task 
for each attribute to a person or group expert in the area. For 
instance, a traffic safety specialist might be called upon to predict 
accident reduction over the null condition or the construction section 
may supply construction cost estimates. The decision maker has the 
responsibility of finding the best expett judgment possible within the 
time and cost constraints of the decision. 

Certain tools are available for making rational estimates of 
future performance. The tools range from the Highway Capacity Manual 
which predicts level of service to presentation models which can be 
used to assess non-user and user visual impacts. Each performance mea- 
sure is best predicted with different devices, and it is the duty of the 
expert judge to apply the proper tool. 

Because the performance measures are predicted, rather than mea- 
sured after the fact, a degree of iincertainty exists as to their values. 

E-29 



The uncertainty is a function of the accuracy of the predictive device, 
the tool mentioned above, and can be expressed as a probability distri- 
bution. Whenever sufficient doubt exists as to the experts' predictive 
power the point value estimate should be discarded in favor of a distri- 
bution. The widths ox ranges of these distributions increase with an 
increase in expert uncertainty. Construction costs may be predicted 
with little uncertainty due to the existtince of good historical data, 
but accident predictions at a particular interchange may vary greatly. 

Step 5: Obtain Worth Transformation Functions 

Each alternative has now been rated on a set of performance mea- 
sures either through a point estimate or a distribution of values. In 
order to combine these measures into a single over-all measure for the 
entire facility, the units must be transformed to worth or utility. 
Since the ranges of the performance measurements have been specified, 
the utility or worth functions can be assessed through the methods 
described earlier = Only one transformation function will exist for 
each performance measure, and its range will include the combined range 
of all the alternacivB designs. 

Those performance measures that were estimated directly in a worth 
scale, of course, do not require transformation functions. However, 
the advantages of estimating performance measures and then transforming 
to worth units over direct worth estimation are obvious both from an 
operational standpoint and in the future defense of the decision. 

Step 6; Ge neiate a Num ber of Weighting Schemes 

Different weighting schemes should be devised which reflect the 
diversity of cpinion throughout the affected community. Examples of 

E-30 



different schemes might be (1) a safety conscious scheme, (2) an aesthe- 
tic conscious strategy or (3) a cost-conscious scheme. These would be 
constructed to give heavier weights to areas of safety, aesthetics or 
costs, respectively, so as to give advantages to the alternatives with 
high scores in such attributes. 

One approach to devising the different weighting strategies might 
be to set one schemie which favors each of the second level goals in the 
hierarchy. Taking the goal structure in Figure E-7 as an example, 
this would mean that five separate schemes could be established. The 
weights across each level of the hierarchy should sum to unity, enabling 
individual weights to be computed by multiplying out along the branch. 

To generate alternative weighting schemes one might consider 
different weights on the second- level goals only, leaving the lower- 
level weights constant, but effectively changing the final individual 
weights through multiplication » 

Figure E-8 illustrates this method for using hierarchical structure 
to generate different weighting schemes. The numbers separated by 
slashes are three alternate weighting schemes which are derived from 
giving different weights to the five second-level goals. These lead 
to three sets of final individual weights at the 24 lowest-level goals. 

Step 7; Assign Prior Probabilities to t he Weighting Schemes 

The three weighting systems illustrated in Figure E-8 can be 
more easily understood by considering the different weights on the five 
second-level goals rather than by trying to look only at the 24 lowest- 
level weights. Table E-1 gives the three weighting strategies in 
summary form. The first. Scheme A, gives more weight to user conven- 
ience and only slightly less to sarety- Scheme B is primarily safety and 

E-3i 



GOAL WEIGHTS 



Aesthetical- 
-ly Pleasing 
[.1/.3/.2] 



"GOOD TRANS- 
PORTATION 



Convenient 

■Transport — 

[.4/.1/.2]* 



Safe 

■Transport - 
[.3/. 4/. 2] 



Low travel times [.3] . . 

Free flow [.4] 

Reduce user stress [.3] , 

Low fatalities [.6] . . . 
Low injuries [.3] . . . . 
Low property damage [.1]. 



■User 
[.4] 



-Low noise [.3] . . . . 
-Visually pleasing [.4] 
•-Comfortable ride [.3]. 



pLow noise [.3] 

■Non-Usei — h Visually pleasing [.3] . 

[.6] hLow water pollution [.2] 

'-Low air pollution [.2] . 



Beneficial 
- to the — 
Community 
[.15/.1/.35] 



Econom- 
ically 
[.5] 



Socially- 
[.5] 



P Increase industry [.3] . . . . 
-Decrease unemployment [.4] . . 
-Increase tax base [.2] . . . . 
'-Increase fire protection [.1]. 



-Improve neighborhood [.2] 

-Improve poor & aged mobility [.2]. 

-Improve recreation [.2] 

'-Permit desired growth [.4] . . . . 



Low Cost 
■-Transport — 
[.05/.1/.05] 



[-Construction [.2] 



Operating J User cost [.6] 



t.8] 



-Facility cost [.4] 



FINAL INDI- 
VIDUAL WEIGHTS 

[.12/. 03/. 06] 
[.16/. 04/. 08] 
[.12/. 03/. 06] 



[.18/. 24/. 12] 
[.09/. 12/. 06] 
[.03/. 04/. 02] 



[.012/. 036/. 024] 
[.01 6/. 048/. 032] 
[.012/. 036/. 024] 



".01 8/. 054/. 036' 
" 01 8/. 054/. 036' 
01 2/. 036/. 024" 
012/. 036/. 024" 



.0225/. 015/. 0525] 
.03/. 02/. 07] 
.015/. 01/. 035] 
.0075/. 005/. 01 75] 



.015/. 01/. 035 
.015/. 01/. 035" 
.015/. 01/. 035" 
.03/. 02/. 07] 



[.01/. 02/. 01] 



[.024/. 048/. 024] 
[.016/. 032/. 016] 



Figure E-8. Sample Weighting Procedure 
E-32 



aesthetically-oriented, while Scheme C is intended to give most con- 
sideration to benefiting the community. All of these schemes appeal 
to at least one group of people within the affected society. In assign- 
ing prior probabilities the decision maker must decide how certain he 
is that an individual scheme represents the majority opinion. 

TABLE E-1. 
SAMPLE WEIGHTING SCHEMES 





Conven- 




Aesthe- 








Weight- 


ient 


Safe 


tically 


Beneficial 




Prior 


ing 


Trans- 


Trans- 


Pleas- 


to 


Low 


Probabi- 


Scheme 


port 


port 


ing 


Community 


Cost 


lity 


A 


.4 


.3 


.1 


.15 


.05 


.3 


B 


.1 


.4 


,3 


.1 


.1 


.2 


C 


.2 


.2 


.2 


.35 


.05 


.5 



In Table E-1 the decision maker has determined that he is 50% 
certain that Scheme C, the community benefit scheme, is representative 
of community desires, 30% certain that Scheme A is realistic and 20% 
certain that Scheme B is preferred by the population. These prior 
probability assignments must be made based on public inputs to the 
designer through public hearings, local government, and special interest 
gi^oups in the framework of the existing design process. Perhaps, in 
the future, the accuracy of these weighting scheme probabilities can 
be improved through the application of public opinion gathering devices. 



Step 8; Monte Carlo Sample the Performance Measure Distributions 

The decision maker now has before him a set of attribute perfor- 
mance distributions tox each alternative, means for transforming them 
into worth functions, and a distribution of weighting strategies to 

E-33 



combine the worths of all attributes. He must combine these distribu- 
tions into a single distribution of a single payoff variable. A Monte 
Carlo sampling technique for both performance measure distribution and 
the weighting schemes can be applied. 

Monte Carlo sampling can be\ accomplished by generating a random 
number between zero and 100 and entering on the probability side of 
the cumulative performance measure distribution curve. Where the random 
number crosses the cumulative curve fixes the value of the performance 
measure for that one sample. The technique is the basis for much com- 
puter simulation work and is fully explained in any operations research 
text. Such strategies have also been previously applied in decision- 
making under uncertainty as documented in a paper by Pouliquen written 
for the World Bank (Pouliquen, 1970). 

This technique can be repeated to yield one performance measure 
for each of the attributes, which can subsequently be transformed to a 
worth value. If this procedure is followed 100 times we will, in effect, 
generate 100 interchanges with performance measures following the pre- 
viously specified performance distributions. 

Step 9: Monte Carlo Sample from the Weighting Distribution 

The same type of sampling can be used to choose a weighting 
strategy. After one set of attaribute worth measures are extracted 
from the previous step, a weighting procedure can be chosen randomly 
according to the prior probability distribution. Application of 
such a scheme would result in one payoff point in a distribution of 
points for each alternative. If 100 interchange worth sets were multi- 
plied by 100 weighting schemes, a distribution of payoffs would result 
for each alternative. These curves would form the basis for decision. 

E-34 



The ability to choose different weighting strategies enables the 
decision maker to test the sensitivity of his weights. In public invest-- 
ment decisions where good data on community preference is almost always 
lacking, this is very desirable. If the decision maker can defend his 
choice in a logical manner he can successfully "sell" his decision 
to many unwilling special interest groups. 

Step 10; Produce Payoff Distributions 

The final step before the dfecision is to graph the computed worths 
for each alternative in a cumulative distribution format. An example 
is given in Figure E-9 of three alternatives evaluated under a group 
of equally likely weighting schemes. The decision maker is presented 
with much more information than a simple mean payoff. (Although means 
could easily be computed and plotted.) The distribution of payoff 
is a much more meaningful device or tool for evaluating alternatives 
since it gives ranges and the shape of the entire function. 

In the example, the decision maker sees that Alternative 2 has 
the highest payoff most of the time or is the best alternate in about 
55% of the simulated cases. In addition. Alternative 2 is the worst 
of the three about 20% of the time. Alternative 1, however, is the 
best selection 45% of the time and is only slightly second best the 
remaining 55%. Also, Alternative 1 is always better than Alternative 3, 

A further refinement of the payoff distribution might be to Include 
the weighting schemes on the graph so that the decision maker could see 
the effect of different weighting schemes on the final payoff. Figure 
E-10 demonstrates this type of information by presenting two alternatives 
which are evaluated using three different weighting schemes. The bands 
represent the outer limits of each alternative as defined by the 

E-35 



I.0-- 



o 



< 

s 

a: 
o 

z 
< 

I- 

cn 
w 
iij 

_i 

i 



>- 

m 
< 
ffl 



75-- 



ALTERNATIVE 2 



.50-- 



^ .45- 



25-- 



.20- — 




PAYOFF (P) 



Figure E-9. Sample Payoff Distributions 
E-36 



Individual weighting strategies. Figure E-10 shows that Alternative 2 
la generally more sensitive to the different weighting schemes since 
the band widths of payoffs are larger. 

Example Problem 

One way to demonstrate the methodology proposed in this appendix 
Is to work through an example problem. The example evaluation will be 
directed toward the choice between two hypothetical interchange con-r 
figureatlons in a very simplified setting. Each alternative will be 
evaluated using three different weighting schemes and 24 separate per"- 
foBmance measures. The example will follow the atep-by-step procedure 
outlined in the previous section. 

Step 1; Establish a Goal Hierarchy 

For simplicity, the example goal hierarchy in Figure E-7 will be 
used. 

Step 2: Establish Performance Measures 

The 24 lowest- level goals and the attendant performance measures 
§j:e shown in the left columns in Table E-2. 

Step 3t Generate Alternative Designs 

It will be assumed that both alternatives are feasible and repre- 
sentative of all reasonable solutions to the problem. 

Step 4; Obtain Performance Distributions 

It is assumed that the set of 24 performance distributions shown 
in Figure E-11 apply to both alternatives. No attempt was made to 
insure that the distributions are illustrative of two real world alterna- 
tives; rather, the distributions are intended to demonstrate the variety 

E-37 



1.0 ^_. 



Ql 

o 



< 

s 

O 



< 

X 



CO 
CO 
UJ 



CO 

< 
o 

X 



>- 



m 

< 
m 
o 

a. 



.75 __ 



.50 __ 



.25 _- 




WEIGHTING A 

WEIGHTING B 

WEIGHTING C 



ALTERNATIVE 1 



ALTERNATIVE 2 



PAYOFF (P) 



Figure E-10. Sample Payoff Distributions with Weighting Bands 

E^38 



eg 
I 



W S 

H 5 



4J 

•H 

5t 



CO 



J3, 

bC 
bd 0) 



pq 



in m 

000000-<l-csi-.d-^ovo->3-<-csjOior^mir>mOO^vo 
sOoo^DojvocsjcNcncNfncncNjcNinr^rOiHncnfir^^r-icMrH 
OOOiHOOOOOOOOOOOOOOOOOOOO 



oooooovooovosf>*vovoinooLnooooooofM 
cn-*co«j-(N«jroovi-fomu-ifnfOrHCMiHOf-iiHrHc>jpg-*fn 

OOOCNtHOOOOOOOOOOOOOOOOOOO 



to 



oooooo(sivo«NoooocNjcNCNjomr>.inmLnoo«*so 

CNvOCNOOCTifni-liHfHiHiHHiHCMnrHOiHrHiHfnrHfsJrH 
iHiHiHiHOOOOOOOOOOOOOOOOOOOO 



H 
O 



u 



W 

(1) M 

Is 

•H W 

^3 o 
(U 

O t-l 

S3 <U 

na > 

P2i iJ 



o «w 

-a 

U 0) 



•n O -W 
•H PL, 



•H Q) 

PI rt 



O 

rH (U 
I 



<u q 






o to 

S CO CO 

(U i-l 

4J C <U 

o ^ ^ 

3 o 

o a> 



&^^ 



o 

U CO 

>.^ 

CO O 

0» -r-; 



<U 0) 

CO CO T3 'O 

nj td <u 

O (U CO 



•H -H 



(1) 
CO 
t« to 



3 
iH 

cd 

> 

•a <u 
<u a 

to -H 

CO 4J 

(U 

CO {>0 

CO > ^ 
n) to -u 
u 
o 

o 

(U 



(U (U 
CO CO 



QpiPPBvS^SMMMOP 



0) 4) 

O O 

3 3 

0) (U 

tfl cfl 

0) (U 

>. >. 

(0 en CO 

>H V4 M 

c4 0) (4 



O 

0) i-J rH 
Vl rH iH 



« o 



















^^^ 












>. 




























u 












4-> 




























0) 












•H 




























m 












rH 
























^^ 




3 










3 


tA 
























^4 /^ 




1 










O 


J2 
























Q) H 




3 










•H 


g 
























CO 0) 




O 






JJ 




JJ 


S 




ja 


S 




•3 














3 CO 


/-X 


3 






3 




O 


•3 




44 


o 




0) 












0) 


w :3 


H 


N-' 3 










(U 


O -3 




jj 


fi 




O 








CO 




60 


•w^ 


0) 


O 






S 




4J 


O 0) 


3 


o 






3 








w 




tO 


t)0 


to 


M-H 


3 


^ 


0) 


o 


J= 60 


O 


Vl 


*i 


"3 


•o 








a> 




S ./-\ 


3 <U 


3 


3 -Ul 


O 


o 


to 


VI 


Vl to 


•H 


60 


to 


U 


(U 




9i 




u 




c3 n 


•H ta 


1 


•H 3 


•H 


4-1 


fH 


to 


o. 


O 


4J 




o 


tJ 


u 




S 




*j CO 




TS OJ 


tn -H 


3 


to .H 


4J 


CO 


t 


rO 




^ <Ji 


CO 


TJ 


o 


3 






•H 




CO Q) 


CO 


CO 


to u 


O 


to rH 


3 


3 




(U 


^ 


<U 


<U 




•3 


u 


iH 


4J 




-H 


(U 


>^ 3 


<u 


3 


(U O 


rH 


•X3 


0) 


X 


Vl 


60 Vl 


V4 


U 


3 


(U 


to 


cd 






V^ -u 


•r) 


4J v-' 


.-1 (U 


Nw' 


rH CU.H 


3 




to 


•w 


•H O 


o 


•r^ 


o 


Vl 


o 


o 


iH 




<U -H 


V4 


M 


&iH 




a. 


O 


•H 


3 


4J 


4H 


0) O 


Q) 


CO 


•H 




o 


o 


(U 


5 


CO H 


•H 


Q) 0) 


JO 


0) 


Kl 


3. 










3 a 


Vl 


0) 


4J 


u 






> 


5 


3 to 


3 


a to 


>, to 


CD 


>^ 0) 




(U 


0) 


(B 


Q) 






•3 


O 


CO 


>» 




to 


fH 


u 


cr 


O -H 


.H 4J 


-H 


rH 4J 


U 


to 


CO 


to 


CO 


0) Q) 


0) 




3 





■u 




^ 


m 


(U to 


3 


M o 


rH U 


O 


.H to 


•H 


to 


to 


to 


to 


> > 


> 


4J 


Vl 


o 


•r^ 




4J 




O l+H 


•H 


ex, 3 


CO O 


3 


to :s 


to 


(U 


lU 


0) 


<u 


o o 


o 


•d 


•M 




rH 






<u 


3 






5'S 




3 




^j 


VI 


Vl 


u 


u u 


Vl 


a 


to 


Vl 


t4 




^ 


(U 


'O ^ 


15 


rs ^ 


:? 


CO S 


15 


CJ 


o 


o 


o 


& & 


i- 


Vl 


3 


4) 


O 




o 


^4 


<u o 


6 


5 5 


•H O 


o 


•H O 


O 


3 


(U 


3 


3 


G 


<u 


o 


CO 


to 




►J 

• 


P:^ 


Pi hJ 

• • 


• 


• • 


• • 


• 


> hJ 


iJ 


M 


• 


M 


• 


M M 

• • 


M 

• 


PM 

• 


• 


• 


• 



cMro-<rmvor^ooo\OiHcgcosi-mvor»ooo\OrH{vjcosf 

rHrHrHrHlHrHrHrHIHrH<Se^JC^^<^^C^^ 



O 
O 
O 



O 

o 
o 



o 
o 
o 



to 

4J 

o 

H 



E-39 



1.0 



(1) 



4-> 



•- .5- 



o 

s- 
a. 



.2 -. 



J 



f 



-i 1 I r- 

5 7 8 9 

Reduced Time (min) 



j-10 1.0 



,8- 



- 6 jc I— -6 - 

4J .r- 

S- J3 

O (O 

[-4 2.4 H 

a. 



h 2 





10 



.2 - 

- 



(2) 



A B C D 

Level of Service 



10 
8 

< 
-4' 

-2 





1.0 



.8 - 



•r= .6- 



o 

Q. 



.2 




2 4 6 

User Stress (DW) 



Alternate 1 
Alternate 2 



1 2 3 

Reduced Fatalities 



Worth 



Figure E-11. Performance and Worth Distributions 

E-40 




Reduced Injuries 



Reduced PDO Accidents 




80 90 
Low Noise (user) (dbA) 



2 4' 6' 8 
Visually Pleasing (user)(DW) 



Alternate 1 
Alterna'te 2 
Worth 



Figure E-11 (cont.) 



^1 




2 4 6 8 10 
Roughness Index 



40 50 60 70 80 10 
Low Noise (non user) (dbA) 



T-IO 
8 
6^ 

4- 
i. 
C 

-2 




(11) 



(12) 




2 4 6 8 10 
Visually Pleasing (non user) (DW) 

• — - — Alternate 1 

Al ternate 2 

Worth 



50 +50 
Water Pollution {% increase) 



Figure E-11 (cont.) 
E-42 




+50% 
Low Air Pollution {% increased) 



-5% +5% 

Increased Industry Payroll {%) 




■6% 




Increased Jobs {% 



+5% 



2% +2% 

Increased Assessment (%) 



Alternate 1 
Alternate 2 
Worth 



Figure E-11 (cont.) 
E-43 




-5 +5 

Decreased Time to Fire (min) 



2 4 6 8 10 
Improve Neighborhood (DW) 



(19) 



(20) 




2 4 6 8 10 

Improved Poor & Aged Mobility (DW) 

. Alternate 1 

Alternate *2 

=^ Worth 



2 4 6 8 10 
Improved Recreation (DW) 



Figure E-11 (cont.) 
E-44 



. ._ 10 




2 4 6 8 10 
Permit Desired Growth (DW) 



$30 mil $35 mil $40 mil 
Construction Cost ($) 




$50K $100K 

User Cost Reduced ($/yr} 

Alternate 1 

— — '-- Alternate 2 

Worth 



-50K +50K 
Facility Cost Reduced {$/yr) 



Figure E-11 (cont.) 
E-45 



of shapes which might be encountered in a typical analysis. Certain 
measures, such as visual pleasure, may be represented as point values, 
while others, such as reduced injury accidents, are shovm to have a 
nearly uniform distribution. The decision maker can represent his esti- 
inate of uncertainty through the shape of his performance measure distri- 
butions. 

Step 5; Obtain Worth Transformation Function s 

The worth transformations for the example are also shox*n in graphi- 
eal form in Figure E-11. As was the case with the performance measures, 
the worth transformations were selected to demonstrate the variety of 
functions available rather than to represent some particular decisipn- 
maker's actual feeling. Worth transforms can be step functions, or 
continuous forms, either sloping positively or negatively. Direct worth 
transforms are simply one-to-one transformations, shown in Figure E^ll 
as a positively sloping 45" line. 

Step 6; Generate a Number of Weighting Schemes 

Three systems for weighting the performance measures are shown in 
Table E-2. These were taken from the goal hierarchy example in Figure 
E-^S and are intended to demonstrate a user convenience oriented system 
(Scheme A), a safety oriented system (Scheme B) , and community-benefits- 
oriented strategy (Scheme C) . 

Step 7: Assign Prior Probabilities to the Weighting Scheme^ 

For the example problem it is assumed that the probability of Scheme 
A being representative of community desires is ,3; Scheme B is .2; and 
Scheme C is .5. This would mean that Scheme C, the community-benefits- 
oriented strategy has the highest likelihood of representing the public's 

wishes, followed by user convenience and then safety. 

E-46 



Step 8: Monte Carlo Sample the Performance Measure Distributions 

A random number between 1 and 100 for each of the 2A performance 
measiures was determined by consulting a table of random numbers. The 
number was used to enter the performance distributions and therefore 
set a performance measure for that particular sample. This measure 
was transformed to worth by using the transformation function for the 
performance measure. The procedure was repeated until a set of worth 
measures for each of the 24 variables were produced for all of the 25 
samples of each alternative. One sample set for each alternative is 
shown below in Table E-3. 

Step 9t Monte Carlo Sample from the Weighting Distribution 

A random number between 1 and 100 was generated for each of the 

50 samples (25 for Alternative 1 and 25 for Alternative 2) . A random 

number less than or equal to 30 Indicated weighting Scheme A, 31 through 

50 Indicated Scheme B, and greater than 50 fixed the weights as C. 

For Alternative 1 there were seven A weights, only two B weights and 

16 C weights. Of the 25 samples of Alternative 2, five were weighted 

according to Schemq A, five with Scheme B, and 15 with Scheme C. 

As the sample sizes are increased the numbers of each weighting 

scheme would approach the expected number,, computed by multiplying 

the prior probability by the sample size. For example, the expected 

number of B weighted samples for Alternatives 1 and 2 is .2 times 25 

or 5. Only two actually appeared in Alternative 1 because of the 

small sample size and the randomized process of selection. For ease 

of hand computation the sample sizes were kept to a minimum, but if 

the procedure were automated, much larger samples (200-300) would be 

appropriate, 

E-47 



TABLE E-3. 
EXAMPLE PROBLEM SAMPLE VALUES FOR ONE SAMPLE 





Performance Measure 


Alternat 
Random No. 


ive 1 
Worth 


Alternat" 


Lve 2 




Random No. 


Worth 


1. 


Reduced time (minutes) 


21 


4 


25 


6 


2. 


Level of Service (A, B, C, 


or D) 84 


3 


40 


6 


3. 


Direct worth 


78 


5 


50 


7 


4. 


Reduced fatal accidents 


38 


3 


77 


7 


5. 


Reduced injury accidents 


18 


2 


48 


7 


6. 


Reduced PDO accidents 


33 


3 


23 


3 


7. 


Decibels at C/L 


63 


3 


96 


1 ' 


8. 


Direct worth 


17 


4 


71 


6 ^ 


9. 


Roughness index (1-10) 


68 


3 


22 


1 


10. 


Decibels at ROW line 


53 


3 


92 


9 


11. 


Direct worth 


12 


8 


26 


fl 


12. 


% increase 


64 


5 


87 


^B 


13. 


% decrease 


25 


3 


55 


^1 


14. 


Increased payroll (%) 


14 


3 


52 


H 


15, 


Increased jobs 


64 


2 


98 


9 W 


16. 


Increased assessed value 


94 


7 


26 


6 m 


17. 


Decreased avg time 


67 


9 


15 


1 


18. 


Direct worth 


65 


8 


74 


6 ; 


19. 


Direct worth 


76 


9 


19 


■7 


20. 


Direct worth 


16 


7 


47 


7 


21. 


Direct worth 


55 


8 


78 


8 


22. 


Dollars 


93 


8 


99 


6 


23. 


Dollars/year reduced 


80 


6 


1 


2 


24. 


Dollars/year reduced 


57 


6 


98 


9 



E-48 



Step 10: Produce Payoff Distr ibutions 

The final step In the evaluacion is to multiply the appropriate weighting 
strategy by the worth values and sum the score to get a total worth. The total 
worth has limiting values of and 10, with the higher scores being preferable. 
The distribution of total worth secies for each alternative is tabulated in 
Table E-4 . From this ranking the cnxvlative distribution of total worth was 
plotted as Figure E-12. The heavy lines axe the worth distributions for each 
alternative taken over all three weighting schemes. 

In addition, Figure E-12 shows the effect of different weighting schemes 
on the total worth of each alternative. These form bands which enable the de- 
cision maker to evaluate each alternative under the best or worst weighting 
condition. 

In comparing Alternatives 1 and 2,. it is obvious that the second alterna- 
tive is preferable. It gr-'es a hij^^ner ^.vijecced utility (total worth score) 
under many sets of circumstances, and in no case does it give a lower one. 
Only in one instance is uhere e tie^ and ':'ha.": occurs when weighting scheme A 
is the true preference of the pub^.iCc Now, .i.i the "unluckiest" 20% of the cases. 
Alternative 2 gives a total worth or caly 4^8, However, we see that under the 
same condition (weighting scheme a) , Aite.r.r.ative 1 may also be that ''unlucky" 20% 
of the time. Furthermore, we aj.so sse cha ' in all outcomes other than the "un- 
luckiest" 20%, Alternative 2 does bcttdX ;.hiv; A.lternative 1 — even in the least 
favorable case (of weighting scheuvfc a) c In an average, 50th percentile outcome, 
it yields a total worth of 5,5 as opposed tc 5.2, Under both other weighting 
schemes, the superiority of Alternaci^'a 2 is even more definitive. 

The foregoing relationship mighc be referred to as "distribution dominance." 
That is, the cumulative distribution or worth scores of Alternative 2 dominates 
Alternative 1 at all probability .^evel3-^ 

This is not the same as saying thit Alternative 2 actually will perform 

better than Alternative 1, however, 

E-49 









TABLE E- 


4 








EXAMPLE PROBLEM TOTAL 


WORTH RANKINGS 








Alternative 


1 


Alternative 2 






Sample 


Total 




Sample 


Total 




Rank 


Number 


Worth 


Weight 


Number 


Worth 


Weight 


1 


21 


3,3090 


B 


15 


4 8585 


A 


2 


8 


3.9750 


B 


7 


5,0350 


B 


3 


11 


4.2155 


C 


25 


5.4965 


A 


4 


16 


4.4370 


C 


3 


5,5730 


A 


5 


1 


4.6000 


C 


5 


5,6505 


C 


6 


17 


4.8585 


C 


9 


5,8170 


c 


7 


2 


4.8625 


A 


22 


5,8725 


c 


8 


12 


4.8875 


A 


17 


6,0285 


c 


9 


24 


4.8880 


C 


1 


6.0650 


c 


10 


10 


4.9205 


C 


16 


6,1015 


c 


11 


9 


4.9495 


C 


14 


6.1395 


c 


12 


13 


4.9865 


A 


21 


6,2070 


B 


13 


15 


5.0420 


C 


24 


6.3510 


B 


14 


25 


5.2015 


C 


2 


6.5040 


C 


15 


7 


5.2180 


C 


13 


6.5785 


C 


16 


4 


5.3385 


A 


8 


6,6045 


C 


17 


22 


5.3790 


C 


12 


6.6210 


C 


18 


5 


5.3815 


C 


23 


6,6290 


C 


19 


3 


5.4735 


c 


20 


6,6800 


A 


20 


23 


5.5575 


A 


18 


6,7275 


A 


21 


20 


5.5630 


c 


4 


6,9635 


C 


22 


18 


5.9650 


A 


11 


6.9845 


C 


23 


6 


6.0680 


C 


6 


7.0965 


c 


24 


19 


6.1285 


A 


19 


7,3210 


B 


25 


14 


6.3520 


C 


10 


7.5250 


B 



E-50 




o 



<: 
I— 
o 



m 

a 
o 

•H 
4J 

•rl 
M 
4J 
CO 

•H 
« 



O 

O 

H 

« 

9< 



«4 



I 

3 
00 
•H 



d > svM HiaoM Ainiavaodd 

E-51 



There may occur individual outcomes in which, due to chance 
factors. Alternative 1 would actually achieve a higher score. What 
is meant here by "chance" factors are (as always) really causal 
elements, but those which must be treated as random variables due to 
our lack of knowledge of how they operate » Clearly, the more that 
such relationships become known and incorporated into the model, 
the fewer erroneous decisions will be generated. Meanwhile, it is 
clearly desirable to recognize explicitly such "chance" sources of 
error. One can thereby not only take the best of calculated risks, but 
also known how likely he is to make a mistake. 

Implementation Difficulties 

It is not likely that the methodology for interchange evaluation 
presented above will be given instantaneous approval by the design 
community. The difficulties which might be expected in implementing 
a decision theory approach can be grouped into user-oriented and 
organization-oriented problems. 

Organizational Difficulties 

An evaluation technique sophisticated enough to be useful in the 

design of major interchanges must rely heavily on a computer. This tie 

to a computer facility, in most cases, means that there must be a 

"middle man" between the designer and the decision. The decision-maker 

must submit his judgment on the various evaluative categories to the 

computer in some format recognizable by each- This coding procedure 

requires transformation of the abstract judgment of a designer to a 

stark numerical form acceptable to the computer. Since most designers 

are not overly familiar with the computer, nor do most computer systems 

E-52 



personnel have an adequate understanding of the design process, a certain 
amount of organizational difficulty is to be expected. 

Highway departments generally lack the organizational structure 
necessary to gather the type of public opinion data necessary to devise 
representative weighting schemes. The public hearing and governmental 
review procedures currently in use are not intended to quantify public 
opinion, nor are the procedures employed really suitable for such data 
collection. The existing process enables those most closely affected 
to voice their dissatisfaction with a proposed design and relies on 
representative government to reflect the wants and needs of the rest 
of the public. Neither of these sources are quantified to the point 
required in a decision theory evaluation. The highway departments, 
therefore, must actively solicit public opinion In a quantifiable manner 
acceptable to the methodological framework. 

A final organizational Impediment to implementing a decision theory 
evaluation In major Interchange decisions is that such a technique 
must be Integrated within the over-all planning process. As was indi- 
cated in Chapter Two, the Interchange design process cannot be separated 
from the over-all transport system planning procedure. Therefore, an 
Interchange evaluation method cannot be applied without integrating It 
Into the larger process. Decision theory evaluative techniques can be 
applied throughout the planning program with only refinements in the 
evaluative categories or subgoals necessary. In addition, such an inte- 
gration would enable the data collection tasks for transport planning 
to be more con5)rehenslve and coordinated. 

User Resistance 

The organizational impediments can be overcome relatively simply 

through the application of additional manpower and financial resources 

E-53 



and some organizational revision. The resistance to a decision theory 
approach by the prospective users cannot be so easily circumvented. 

The designers who make the decisions are by-and-large not familiar 
with the language or concepts of Bayesian decision theory. Rather than 
regarding the techniques as aids to decision-making, they tend to see 
decision theory approaches as "numbers games" aimed at taking the deci- 
sion out of their hands. The prospective users must be educated to 
the advantages of the theory and be persuaded that it doesn't replace 
judgment but rather focuses it. 

This focusing of informal judgment is itself a difficulty in user 
acceptance. To force oneself to quantify judgment could be a painful 
exercise which would deter the user from adopting the evaluative tech- 
nique. This discomfort "cost" is magnified when one is uncertain 
about the benefits of such a technique. The decision-maker can never 
really "know" if the application of the methodology presented above 
will lead him to a sufficiently better decision to justify the real 
and discomfort costs. 

Trade-Off Analyses Extension 

Introduction 

The largest scale problem in interchange evaluation is the evalua- 
tion of alternative configurations. The smaller scale problem — the 
level of investment in the individual components — may be analyzed 
through a scaled down version of the methodology presented above. 
When the problem becomes one of the choice between alternative lengths 
of deceleration lanes, for example, the goal hierarchy and weighting 
schemes can be reduced to a very simplified evaluation system. In 

E-54 



fact, evaluative attributes such as safety, operations and cost may be 
considered as the only attributes with any weight at all in the analysis. 

The problem, then, is reduced to a trade-off analysis where coats 
are balanced against one other attribute, operational and safety bene- 
fits. Therefore, the weighting strategy which could be used to combine 
these two attributes is superfluous to the analysis — direct comparJ(.i- 
sons would be just as meaningful. 

The uncertainty involved in the cost and operational and safety 
performance levels may also be reduced to a point estimate. The cost 
of added pavement in a longer deceleration lane can be accurately esti- 
mated, for example, making point value analyses acceptable. It further 
simplifies the methodology, promoting increased useage. 

This problem approach was discussed at the workshops to obtain 
the designers' views on such a scheme. It should be noted that at 
the time of the workshop the decision theory approach had not been 
devised so that some of the terminology in the trade-off presentation 
was not consistent with the previous section. The most obvious differ- 
ence is between the "Level of Merit" concept introduced in the trade- 
off section and the worth or utility scores discussed in the initial 
part of this appendix. The concepts are essentially the same. 

Workshop Questionnaire 

After some introductory remarks on the need for trade-off analysis, 
a set of discussion questions were posed to the workshop participants. 
These were followed by a period of open discussion and distribution of 
a questionnaire. The participants were asked to complete and return 
the questionnaire the next day — thereby giving them an opportunity to 
discuss the subject further among themselves and to consolidate their 

E-55 



thinking. In general, the questions were nearly the sajne as those 
presented for discussion. The questions, with answers received, are 
given in Table E-5. 

As can be seen from Question #1, the interest in and feeling of 
necessity for economic analyses decreases somewhat as the design deci-^ 
sion becomes more and more specific. This is logical in that the 
alternative costs become relatively smaller and the over-all project 
constraints are pretty well set by the time the design details are 
3eleeted. A number indicated that more economic analyses would be 
desirable but that appropriate methodology was not available. However, 
there is no clear mandate for the development of this methodology. 

Answers to Question #2 indicate that "engineering judgment" is 
the most used decision-making procedure on including "desirable fea- 
tures." It is perhaps surprising, and certainly encouraging, that only 
about a third of the respondents indicated their organisation had 
adopted the policy of simply meeting certain minimums. 

Again, in Question #4, it is apparent that experience is the pirime 
input to the design decision process, although considerable attention 
is being paid to accident record analyses and pertinent research results. 

jCievel of Merit Concept 

Cost and some measure of operations and safety are two major tr^d^T 
pff factors receiving consideration in the selection of alternative comr 
ponent configurations (such as left vs. right ramps, single vs. double 
exits, etc.) and in the specification of design dimensions (design speed 
for a given ramp, length of acceleration lane in a given situation, etc.). 
In development of a final interchange design, a number of these trade- 
off decisions are made (although, perhaps, not "consciously"), 

E-56 



^ 



TABLE E-5. 
RESULTS FROM QUESTIONNAIRE ON TRADE-OFF ANALYSIS 



Questions and Answer Choices 

1, Economic analyses (cost/benefit ratios, 
rate-of-return methods, etc.) as applied 
to major interchange design. Please 
circle the statement you feel most appro- 
priate: 

a. Economic analyses in comparing 

alternative interchange configura- 
tions as a whole 

i. Common practice 

ii. Desirable and feasible, but not 
usually carried out 

iii. Desirable, but not feasible — 

appropriate methodology not vavail- 
able 



No. of Participants 
Selecting Given 
Answer 



iv. Of little practical value; other 
considerations are determining 
factors 

v. Other 



b. Economic analyses in selection of alternative 
components — (loop ramp vs. direct connec- 
tion; collector-distributor roadway vs. 
double exit, etc.) 

i. Common practice 

ii. Desirable and feasible, but not 
usually carried out 

iii. Desirable, but not feasible — 

appropriate methodology not avail- 
able 



11 
6 



iv. Of little practical value; other 
considerations are determining 
factors 

V. Other 



12 
5 



E-57 



Table E-5. (Continued) 



Questions and Answer Choices 

c. Economic analyses in specification of 
design dimensions — (length of accel- 
eration lane, radius of curvature of 
loop raxapf etc.) 

i. Common practice 

ii. Desirable and feasible, but not 
usually carried out 

iii. Desirable, but not feasible — 

appropriate methodology not avail-- 
able 

iv. Of little practical value; other 
considerations are determining 
factors 

V. Other 



No. of Partici- 
pants Selecting 
Given Answer 



2. How do you reach decisions on "desirable features," 
such as exclusion of left-hand exits, good visi- 
bility of the exit area, uniformity of exiting 
maneuvers, etc.? (Circle one) 

a. Decision to meet AASHO Blue Book minimums 
at all costs 

b. Decision not to Incorporate (or exclude) 
certain features at all costs 

c. Attempt benefit/cost (or similar) analysis 
for individual situations 



17 
2 



d. Engineering judgment — i.e., no formal 
analysis of cost factors as such 

e. Ocher 



15 
1 



E-58 



Table E-5. (Continued) 

No. of Partici- 
pants Selecting 
Questions and Answer Choices Given Answer 

3. Can meaningful cost data be obtained for in- 
dividual components (ramp configurations, 
length of deceleration lane, etc.)? 

a. Yes; Comment 19 

b. No; Comment 9 

4. How do you assess "benefits" to justify extra 
expenditures for improving on "minimum" design 
standards? (Circle any appropriate answers) 

a. Accident record analyses of similar situations 15 

b. Experience in observing similar situations, 

and relating this to extra costs involved 19 

c. Study of research results in these areas 12 

d. Consensus of personnel in your design 

department 12 

e. Usually use minimum values 

f. Other 7 



E-59 



If design engineers are asked: — "Is it more desirable, from an 
operations and safety viewpoint , to provide a single exit (with sub- 
sequent branching for left and right movements) or two individual 
exits?" — The answer is almost unanimously: — "Single." However, if 
then asked which configuration should be established as a design standard 
to be rigidly adhered to, the answer becomes somewhat less definite, 
and "hedging" will be noted. Obviously, the hedging comes about because 
designers feel there are "situations" in which the single exit should 
not be selected; and this is often because, in that situation, the 
double exit could be achieved at considerably less cost. 

The same types of questions and answers can be applied to other 
design features, such as right vs. left ramp, length of acceleration 
lane, etc. In other words, there are known desirable features, but 
something less is often used because of some cost factor. Designers 
claim it is impossible to give a set answer to any of these types of 
questions which will hold across all situations. A major reason for 
this is that they are trying to assess cost and merit (worth or utility) 
measures at the same time, and, as the combinations are nearly infinite, 
so are the "correct answers." 

It appears, then, that since no definite universal answers can 
be had when the two factors are considered together, it x^/ould be helpful 
to decision-makers if they could assess the two factors (cost, and 
operations and safety merit) individually and then make their decision 
on the basis of relative costs and relative merits. 

Assessing relative costs will usually be possible, though some- 
times with considerable difficulty if the alternatives are such that 
a major portion of the interchange design is involved (such as a deci- 
sion on a right-hand or a left-hand exit) . In the case of designating 

E-60 



I 



the length of an acceleration lane, the cost analysis may be vexy 
simple (if only a little change in earthwork quantities and pavement 
length is required) or somewhat more difficult if the longer lane will 
also interfere with downstream features, require a larger grade separa- 
tion structure, etc. 

The problem, then, will be to assess the relative level of merit 
provided by the alternative configurations, or the alternative design 
dimensions, and then to choose among the alternative levels of per- 
formance and the corresponding costs. 

Assuming, for the moment, that the specification of alternative 
merits is possible, the designer is then in a much better position to 
select the final design. This will still be a highly subjective pro- 
cess, dependent largely on the designer's engineering experience and 
judgment — a "benefit-cost analysis" is not being suggested. 

An example will illustrate the concept. Assume the following 
conditions: 

Configuration Merit Rating Add itional Cost 
Single Exit (on right) 10 $3,000,000 

Double Exit (both right) 8 $2,000,000 

Double Exit (right & left) 3 

. If the total interchange cost (with double exit, right and 
left) is estimated at $40,000,000, which conf iguration^ should 
be selected? 

. If the total interchange cost (with double exit, right and 
left) is estimated at $7,000,000, which configuration should 
be selected? 

. Now assume the ratings are changed to 10, 8, 6: — Which con- 
figuration should be selected? 

E-61 



The fact that different configurations might be chosen under these 
differing conditions points up the problem of setting definitive con- 
figuration selection criteria. Even in this simple example (in practice, 
other considerations, such as maintenance costs, road user costs, etc., 
would also enter the decision-making process) it is not possible to 
select a single "always correct" answer. 

The merit ratings give some insight to the question of "How much 
better?" It is agreed that a single exit is better than one Incorporating 
a right-hand and left-hand exit, and therefore using a design incorporat- 
ing a single exit justifies a higher cost — but how much higher? First, 
one must determine how much "better" one configuration is than another. 
The merit ratings, if available, could provide some "feel" for these 
qualitative comparisons. 

Each time a decision has been made in the past, the designer did 
go through some similar assessment of the relative merits and costs. 
The merit ratings, If they can be developed in a credible and acceptable 

manner, will provide some basis for a rational choice. They would 
provide a means by which the decisions could be made more consistently 
by each designer, and more consistent designs could be obtained from 
various designers. 

As another example, assume a speed change lane (acceleration) from 
a turning roadway with a design speed of 40 mph to a through roadway 
with a design speed of 70 mph must be designed. The "Blue Book" sug- 
gests this acceleration lane should be 1,000 ft. long. Suppose, due 
to situational considerations, a speed change lane 800 ft. long would 
be $500,000 less expensive than one 1,000 ft. long; which should be 
selected? 



E-^2 



Obviously, a Judgment on the importance of that missing 200 ft. 
is required. This assessment is usually made on the judgment of the 
design engineer. Suppose, however, that credible "merit" ratings are 
available — "8" for 1.000 ft., and "7.5" for 800 ft. Wouldn't this 
affect the decision in a different manner than if the two ratings were 
"8" and "A"? Wouldn't this degree of specificity help the designer 
in making this decision? 

I llustrative Rating Questionnaire 

At this point in the workshop discussions, the participants were 
ask^d to complete the questionnaire shown in this section, to illus- 
trate the feasibility, and problems, of deriving consensus "expert" 
judgmental evaluations. 

"The procedure in filling out this questionnaire is quite simple; 
set a value of '10' for the most desirable alternative presented, and 
then rate the others against that one on the basis of operations and 
safety -- keeping In mind that '0' designates totally unacceptable. 
Costs will be considered later in the decision-making process, and 
are not to be a factor here." 

1. Various possible exit ramp configurations for an approach to a 
major interchange at the crossing (roughly perpendicular) of two freeways 
axe shown in Figure E-13. Assume single-lane turning roadways, four-lane 
freeways and that the CHV for each turning movement is 1,000 vph. 



E-63 



® 



® 





© 



<? Lancs- 





® 



(D 



Z LfiN€s —J 




Z LaN€5 



ZL 




© 



2 Lamps 




Figure E-13. Alternative Exit Ramp Configurations 

E-64 



y^S^^^ Rating (Operations and Safety ]) 

A 



B 
C 
D 



2. Alternative lengths for an acceleration lane of a major Inter- 
changa. Turning roadway (single lane) design speed Is 40 mph, through 
roadvay Is 70 a^h. (Blue Book value Is 1,000 ft.) Assume DHV of 1,000 
vph. 

Length (ft.) Rating (Operations and Safety) 
1400 



1200 
1000 
800 
600 
400 
200 




E-65 



Results from Illustrative Rating Questionnaire 

A tabulation of the ratings given to the laternatlve ramp configura- 
tions Is given in Table E-6. In tabulating the results, the parti- 
cipants were "categorized" Into three groups — Design Engineers, 
Traffic Operations Specialists, and Academic and Research — so that 
any differences of opinion among these three areas of expertise could 
be noted. The number of participants in each group returning the 
questionnaire is Indicated at the bottom of the Table. 

It Is interesting to note that all three groups select configura- 
tion E as the "best," and consider the left-hand exit designs the least 

TABLE E-6 
MEDIAN MERIT RATINGS FOR EXIT RAMP CONFIGURATIONS 



Merit Ratings (Median of those responding) 



Figure 


Design 
Engineers 


Traffic Operations 
Specialists 


Academic 
& Research 


All 
Groups 


A 


1 




1 


3 


1 


B 


2 




2 


3 


2 


C 


6 




4 


4 


5 


D 


8 




8 


6 


8 


E 


10 




10 


10 


10 


P 


6 




6 


6 


6 


Number 
Responding 


18 




6 


7 


31 



E-66 



desirable. The Traffic Operations Specialists gave slightly lower rat- 
ings to the loop ramp configuration (C) than did the Design Engineers. 
Although the sample Is small, the results tend to Indicate that those 
who work with the "product" on a day-to-day basis feel even more effort 
(and money) should be expended to eliminate "second-choice" design 
features. 

In general, those categorized as Academic and Research were not 
quite as critical of the left-hand exit designs as the other groups. 
A possible Interpretation Is that the Academic and Research group base 
their opinions primarily on conceptual principles and that. In fact, 
actual operations and safety at left-hand exit ramps are even poorer 
than might be anticipated. 

The results of the ratings of the alternative lengths of accel- 
eration lanes are shown In Table E-7. 

Again, It can be noted that the three groups are essentially In 
agreement, with the Design Engineers being slightly less critical of 
"sub-standard" design. 

It is also interesting to note that the Blue Book value has a 
meclian rating of "9" — indicating that the participants believe this 
value to be adequate. A slightly higher value is reported for 1200 ft., 
but then it tends to drop off again as the length is extended further. 
From comments, it would seem this dropping off is due to concern for 
the excessively long merging area which might result, or the possibility 
that drivers might temporarily believe, the lane was not going to be 
dropped . 



E-67 



TABLE E-7 
MEDIAN MERIT RATINGS FOR ACCELERATION LANE LEHGHTS 

Median Ratings (Average of those responding) 



Length (ft.) 


Design 
Engineers 


Traffic Operations 
Specialists 


Academic 
and Research 


All 
Groups 


1400 


10 


9 




9 


10 


1200 


10 


10 




10 


10 


1000 


9 


9 




8 


9 


800 


7 


6 




4 


6 


600 


3 


1 







1 


400 
















200 

































Number 
Responding 


18 


6 




7 


3X 



The use of group medians In Tables E-6 and E-7 masks the rather wide 
range of Individual ratings, as the "outliers" are lost in this pro- 
cess. As examples, the ratings for configuration A in Figure 1 ranged 
from "0" to "7"; configuration D from "3" to "10"; and configuration E 
from "7" to "10." These large discrepancies may indicate an interpre- 
tation problem on the part of some of the respondents, or differences 
in past experiences with the various designers. Hence, the use of 
the "Delphi Method," as described by Dalkey and Helmer In an article 
entitled "An Experimental Application of the Delphi Method to the 
Use of Exports" ( Management Science , vol. 9, 1963), or some similar 
technique for arriving at concensus opinion Is suggested for future 
studies of this type. 

E-68 



Further Introductory Remarks 

Before beginning the open discussion In the workshop , It was 
further pointed out that If these merit ratings can be set for alter- 
native configuration choices and for design dimensions, the possibility 
for specifying different "levels of merit" (or total worth) for entlr« 
Interchanges exists. For example, for a major Interchange, the designer 
could specify that all configurations and design dimensions must have 
merit ratings of "9" or better; while for a less Important Interchange, 
configurations and dimensions with ratings of "7" might be acceptable. 

Hence, these merit ratings could be used to select Individual 
design featxires through comparison of relative merits and relative 
costs, or as a means to assure design features consistent with the* 
"importance" of the interchange, and, if desirable, consistent within 
A given Interchange. 

This last statement leads to another question: — "Is it ever 
desirable to purposely degrade a design feature so that the 'level of 
design' will appear to be consistent to the driver?" In other words, 
is it better if the driver encounters marginal quality throughout the 
interchange than if he observes high quality in all places in the 
interchange except at one critical site? Will he be deceived into 
thinking he la on a better grade facility than he is, in fact? 

Questionnaire Results 

In addition to the illustrative rating questionnaire handed out 
during the introductory remarks, a session questionnaire was given to 



E-69 



T/JBLE E-8 
RESULTS FROM QUESTIONNAIRE ON LEVEL-OF-MERIT DESIGN CONCEPT 



Questions and Answer Choices 

1. Do you feel It Is possible to derive mean- 
ingful ratings for alternative general 
configurations (as In the example of the 
various exit ramp configurations)? 



a. Yes; Comment_ 

b. No; Comment 



2. Do you feel It Is possible to derive meaning- 
ful ratings for alternative design dimensions 
(as In the example of the acceleration lane 
lengths) ? 



a. Yes; Comment_ 

b . No ; Comment 



3. How should the merit ratings be developed — 
utilizing which Inputs? (Circle all you feel 
apply.) 

a. Physical analyses (acceleration potentials, 
friction factors, reaction times, etc.) 

b. Accident data across alternatives 

c. Research studies on driver behavior and 
preferences 

d. Judgment of highway designers and opera- 
tions specialists 

e . Others 

4. Were you "comfortable" making the ratings 
requested in the earlier examples? 



a. Yes; Comment_ 

b. No. 



No. of Partici- 
pants Selecting 
Given Answers 



18 
12 



19 
11 



20 
20 

21 

17 
8 

17 
12 



If no, what additional information would have 
been helpful? 



E-70 



Table E-8. (Continued) 



No. of Partici- 
pants Selecting 
Questions and Answer Qioices Given Answers 

5. Do you feel the concept of using level- 
of-merit ratings in interchange design 
is: 

a. feasible? Yes No Comment Yes - 14 No - 6 

b. practical? Yes No Comment Yes - 5 No - 11 

c. deserving of more investigation, better 
definition, more trial, etc.? 

Yes No Comment Yes - 17 No - 4 

6. Is consistency in interchange "quality" 
important? Should some elements purposely 
be degraded to make them compatible with 
the lower standard design-controlling 
elements? 

a. Yes, usually. Comment 5 

b. Yes, sometimes. Comment 9 

c. No. Comment 13 



E-71 



the participants at the end of the discussion, and they were asked to 
complete it and return it the following day. As in the case of the 
questionnaire on Trade-Off Analyses, the questions generally paralleled 
those employed to structure the discussion. The questions, with tab- 
ulations of the answers, are given in Table E-8. 

The answers to Questions llll and ill indicate somewhat more than half 
the participants believe it is possible to derive meaningful merit 
ratings. The Design Engineer group was about evenly split, \^ile the 
other two groups were considerably more optimistic. 

The results of Question #3 are not very informative, in that vir- 
tually everyone felt that all possible Inputs should be utilized in 
developing these merit ratings (assuming they should be developed) 

In Question H, a number of the participants indicated they were 
"not comfortable" making the ratings, but they provided little infor- 
mation as to what would have been helpful. (The signing and lighting 
conditions were mentioned as other possible information Inputs.) 

From the results of Question #5, it can be seen that the parti- 
cipants generally felt the level-of-merit design worthy of more in- 
vestigation and trial, but were not optimistic about obtaining a 
practical design tool. 

No clear-cut conclusion can be drawn from the answers to Question 
^6, This is perhaps due to the wording of the question — the comments 
accompanying the answers indicated that the participants were inter- 
preting this question in a variety of ways. 



E-72 



Conclusions 

The value of the evaluative methodology presented above measured 
In terns of Improved decisions Is unknovn. Due to the complexity and 
magnitude of the problem It Is Infeaslble to design an experiment which 
could establish the benefits expected from the use of a decision 
theory approach. The best argument for Its adoption la simply that 
similar techniques are being used currently by large corporations with 
apparent success and the perceived trend is toward expansion of their 
uss. 

There will be problems in implementing the technique within the 
highway decision framework — both with organizational structure and 
Individual resistance. Highway departments are generally not organized 
as a corporation in that some of the staff services groups required for 
data collection and decision theory analysis do not exist. The decision- 
makers themselves are not generally acquainted with decision theory con- 
cepts nor are new engineers being trained in this particular field as 
business school graduates are. Remedial education is needed at the 
decision-making level and program revision required in current engineer- 
ing training before the method will be applied on any significant- scale. 

If the decision theory approach is adopted and executed properly 
It will force Informal evaluation to be better focused, which will lead 
to a better understanding of the reasoning process. It makes the hidden 
assunq>tlons Inherent in the decision explicit and, therefore, subject to 
scrutiny. Both of these characteristics of decision theory approaches 
are desirable if one accepts the notion that the more we know about a 
process, the better the process will work. 



E-73 



Finally » decision theory analysis can be an effective communication 
vehicle for conveying the analysis underlying the choice of a particular 
Interchange design* This Is necessary to convince both the public and 
design reviewers of the appropriateness of a particular design. Such 
a presentation vehicle would fill an existing void. 



< 



i 



* 



E-74 



APPENDIX F 



ACCIDENT ANALYSIS OF MAJOR INTERCHANGES 



F-i 



^i 



CONTENTS 

Page 

Introduction . o.,,, .0=. c = .-,,=. oo, ... o ,. . F-1 

Major vs. Minor Incerchanges .... o o., 00 o ..... . F-2 

Accident Locations on Major Interchanges .. c ....... . F-4 

Accidents for Different Connections on 

Major Interchanges .,. ...o .... .<.,....,. . F-7 



F-ii 



TABLES 



Page 



F-1 Accident Characteristics of Major and 

Minor Interchanges «,.=.., F-2 

F-2 Injuries par Accident for Interchange Types » . » . . . . F-3 

F-3 Acceleracion vs. Deceleration Area 

Accidents on Major Interchanges , . . . o F-5 

F-4 Left Side vso Right Side Acceleration 
and Deceleration Area Accidents 
on Major Interchanges ........ o .., o .. . F-5 

F-5 Accidents by Type of Terminal on 

Major Interchanges .......... o . F-6 

F-6 Accidents by Type of Connection on 

Major Interchanges o .... . F-7 



F-iii 



APPENDIX F: ACCIDENT ANALYSIS OF MAJOR INTERCHANGES 

Introduction 
This limited analysis of major Interchange accident characteristics 
attempts to answer three basic questions regarding interchange operation. 

(1) How do the accident patterns at major interchanges 
differ from those at *minor interchanges? 

(2) Where on major Interchanges do accidents most frequently 
occur? 

(3) What types of connections are the least hazardous; loop, 
direct or semi-direct, or outer connection? 

The data base used for the accident analyses is an automated acci- 
dent record system maintained in the FHWA Interstate System Accident 
Research Program. The information system allows the user to obtain tabu- 
lations of various response variables, crossed with numerous geometric, 
traffic operation, and environmental characteristics. The information 
reported below was supplied to this research agency (via the Contract 
Manager) by Ms. Julie Fee, Head of the Interstate Accident Study group. 

As with many studies where the original data has been collected for 
another purpose by another agency, many of the specific data requests 
could not be fulfilled. Thus the data shown here represent a compromise 
between what was requested and what could be supplied. Consequently, 
this discussion does not constitute a comprehensive study of major inter- 
change accident characteristics. 

The data which were provided represent accident histories for 1,688 

minor and 37 major interchanges. The imbalance between the two types 

of interchanges Implies a great deal about the confidence one may place 

in the resulting figures. By virtue of the significantly larger numbers, 

F-1 



the statistics on minor interchanges are necessarily more reliable, and 
therefore can be more easily generalized. With only 37 major inter- 
changes, representing a wide range of geometries, within the data base, 
one must be considerably more cautious in generalizing from the resulting 
average figures. 

Maj or vs . Minor Interchanges 

The initial question posed dealt with a comparison of the general 
safety characteristics of major versus minor interchanges. Table F-1 
presents the accident, injury and fatality rates (per 100 million vehicle 
miles) by rural and urban types of major and minor interchanges. 

TABLE F-1 
ACCIDENT CHARACTERISTICS OF MAJOR AND MINOR INTERCHANGES 

Interchange Type 



Minor 


Maj 


or 


Urban 


Rural 


Urban 


Rural 


240 


122 


174 


225 


155 


81 


112 


149 


3.1 


4.3 


2.9 


4.1 


529 


1059 


15 


22 



Accident Rate 
(Accidents/100 M veh-mi) 

Injury Rate 
(Injuries/100 M veh-mi) 

Fatality Rate 
(Fatalities/100 M veh-mi) 

Number of Interchanges 



Accident rates for major and minor interchanges have an irregular 
pattern when stratified by urban and rural locations. That is, accident 
rates at major interchanges are lower in urban areas than in rural areas, 
whereas the opposite is true for minor interchanges. 

F-2 



One might assume that since turning volumes on major interchanges 
are usually much heavier than on minor ones, the major interchanges 
would be more hazardous due to the many merges and diverges. The results 
support this assumption except for the urban-minor case, where the rate 
is the highest of all four interchange classes. The explanation for 
this high rate may lie in the types of ramp connections with the cross 
streets. These would typically be diamond connections with high volume, 
at-grade junctions controlled by signals or stop signs. Since accidents 
in this area may have been classified as "interchange" accidents, this one 
feature inflates the accident rate artifically for the urban-minor inter- 
change category. Extremely low cross street volumes would decrease the 
hazard at rural-minor locations. 

The injury rate statistics reflect the same rankings as the acci- 
dent rate figures. A measure of the severity of the accidents which 
occur on each type of interchange would be the injuries per accident, 
which can be computed by dividing the injury rate by the accident rate. 
The results of this calculation are shown in Table F-2, It is obvious 
that the severity of accidents, as measured by injuries per accident, 
are essentially the same for all interchange types. 

TABLE F-2 
INJURIES PER ACCIDENT FOR INTERCEANGE TYPES 

Interchange Type 
Minor Major 

Urban Rural Urban Rural 
Injuries /Accident ' .65 .66 ,65 .66 



F-3 



A,ccident Locations on Major Interchanges 

The second question posed is where on major interchanges do acci- 
dents occur and where are they most severe. Table ^-3 shows accidents 
per deceleration area and per acceleration area. On major interchanges 
it appears that deceleration areas are more hazardous than acceleration 
areas. The term "area" is used here rather than lane because the data 
for some sites include the appropriate half of a combined accel-decel 
lane. 

The number of units shown in Table F-4 and subsequent tables does 
not refer to the actual number of sites in the data base. It refers, 
instead, to the number of location entries in the data base. Therefore, 
some double counting occurs when the same location is entered into the 
file for two separate time periods. This accounting method requires 
that extreme caution be used in interpreting these data. 

Table F-4 is refined from the previous table in that left- and right- 
side subdivisions have been added. The first thing to note in Table 
F-4 is that only three left-side acceleration areas are represented in 
the sample, rendering the per unit statistics virtually useless. There 
are 26 left-side deceleration areas which, while far from the 330 right- 
side total, are enough to make some limited inferences on the safety of 
left-side terminals e The accident rate per unit is nearly twice as high 
for left deceleration areas as for the right side. 

Table F-5, showing accidents by terminal type and location, is the 
most finely divided. It shows deceleration and acceleration lanes 
separated from combined accel-decel lanes and includes left- and right- 
side breakdowns. The table indicates that combined lanes operate more 
safely than separate acceleration or deceleration lanes. 

F-4 



TABLE F-3 
ACCELERATION VS. DECELERATION AREA ACCIDENTS ON MAJOR INTERCHANGES 



Unit 



Number of 
Units 



Nuiober of 
Accidents 



Accidents/ 
Unit 



Deceleration 
Area 

Acceleration 
Area 



356 



323 



431 



271 



1.21 



.84 



TABLE F-4 

LEFT-SIDE VS. RIGHT-SIDE ACCELERATION AND DECELEMTION 
AREA ACCIDENTS ON MAJOR INTERCHANGES 





Number of 


Number 


of 


Accidents/ 


Unit 


Units 


Accidents 


Unit 


Deceleration 










Right Side 


330 


376 




1.14 


Left Side 


26 


55 




2.12 


Acceleration 










Right Side 


320 


243 




.76 


Left Side 


3 


28 




9.33 



F-5 



TABLE F-5 
ACCIDENTS BY TYPE OF TERMINAL ON MAJOR INTERCHANGES 





Number of 


Niimber of 


Accidents/ 


Unit. 


Units 


Accidents 


Unit 


Deceleration Lane 








a. Right Side 


242 


311 


1.29 


b. Left Side 


23 


50 


2.16 


c. Combined 


265 


361 


1.36 


Acceleration Lane 








a. Right Side 


235 


185 


.79 


b. Left Side 


3 


28 


9.33 


c. Combined 


238 


213 


.89 


Deceleration Half 








of Combined 








a. Right Side 


88 


65 


.74 


b. Left Side 


3 


5 


1.67 


Ct Combined 


91 


70 


.77 


Acceleration Half 








of Combined 








a. Right Side 


85 


58 


.68 


b. Left Side 








None 


c. Combined 


85 


58 


.68 



F-6 



Accidents for Different Connections on Major Interchanges 

The final question considered in this analysis is that of the 
relative safety of different types of connections; specifically loop 
ramps, outer connections, and direct or semi-direct turning ramps. 
Table F-6 contains accident rates expressed as the number of occurrences 
per 100 million vehicle-miles for the three types of connections, for 
both one-t. and two-lane widths. 

TABLE F-6 
ACCIDENTS BY TYPE OF CONNECTION ON MAJOR INTERCHANGES 



Number of Accidents/ 

Unit Units 100 M veh.-ml, 

Loop Ramp 

a. One Lane 42 355 

b. Two Lanes 40 485 

Outer Connector 

a. One Lane 22 361 

b. Two Lanes 66 171 

Direct or Semi- 
Direct Ramp 

a. One Lane 11 98 

b. Two Lanes 21 164 



The rates shown in Figure F-6 should be interpreted with caution. 
The number of units for any category is relatively small, particularly 
in view of the possible double counting involved in the unit tabulation. 
The small number of units for each category probably explains the 
apparently Irrational relationships among the accident rates. 



F-7 



For example, one would not expect that one-lane outer connections 
would be much more hazardous than one-lane direct or semi-direct ramps; 
yet the rates indicate that the outer connection ramps are four times 
more dangerous. Further, there is no apparent explanation for the 
indication that one-lane outer connections are twice as dangerous 
as two-lane outer connections. 

For these reasons, no specific conclusions are drawn from the data 
in Table F-6. 



F-8 



APPENDIX G 
EXIT TERMINAL CASE STUDY 



J 



William J. Laubach, The Pennsylvania State University 

G-i 



CONTENTS 



O 



Site Description ..„...„, o =» o o » 
Critical Analysis of Existing Design Features 

Deceleration Lane Shape 

Left-Hand Exit , 

Single Versus Double Exit Design 
Suggested Design Improvements 



t 4 n 



e a 



o e o 



Page 

G-1 

G-3 

G-3 

G-9 

G-22 

G-23 



G-ii 



FIGURES 

Page 

G-1 Schematic of Interstate 80 and 81 Interchange „ . , . G-2 

G-2 Projected Traffic Volumes (VPH) , . . » „ „ » » , . „ G-4 

G-3 Northbound Approach to Exit Ramp H „,„»,,,,. G-6 

G-4 Southbound Approach to Ramp F ...,.„..',,, . G-8 

G-5 Existing Deceleration Lane for Ramp A . „ . , o = » » G-10 

G-6 Existing Deceleration Lane for Ramp D o »,,,,, . G-11 

G-7 Alignment for 180 Eastbound . . . o » « . . « o , ■, . G-14 

G-8 The Existing Signing for the Left-Hand Exit , » . „ , G-16 

G-9 Suggested Signing Plan for Left-Hand Exits , , . . , G-19 

G-10 Recommended Supplementary Signs , » » . , , . o ^ . » G-20 

TABLES 

Page 

G-1 Deceleration Lane Shape and Length ,, o o .<.<= = . G-5 

G-2 A Comparison of the Existing and Recommended 

Distances between Successive Exits , , . » , c . , . G-24 



G-iii 



APPENDIX G: EXIT TERMINAL CASE STUDY 

The following critical analysis of the exit terminals of an exist- 
ing major interchange provides an example of how the design recommenda- 
tions and conditions of the previous chapters can be applied to analyze 
or critique an existing design. 

The data used here were collected during visits to the interchange 
site, the state highway district office, and the office of the consulting 
engineers who designed the interchange, 

S it e De s crip t ion 

The major interchange chosen for analysis is the interchange between 
Interstate 80 (Legislative Route 1009) and Interstate 81 (Legislative 
Route 1005) lying in a rural area of Luzerne County, Pennsylvania. Inter- 
state 80 is the major east-west route across northern Pennsylvania, and 
Interstate 81 is a major north-south route in the eastern half of the 
state. The major destinations from the interchange are Scranton and Wilkes 
Barre to the north, Harrisburg to the south, Bloomsburg to the west, and 
Stroudsburg to the east. 

A schematic of the interchange is shown in Figure G-1. The inter- 
change is relatively new, having been designed in 1963- Basically, the 
interchange can be described as being a modified clover leaf. Three of the 
four left turn movements are accommodated by loop ramps. The remaining 
left turn, from east to north, is served by a left-hand exit. The design 
speed for both freeways is 70 miles per hour, and the posted speed limits 
are 65 miles per hour. Both freeways are four-lane facilities, and all 
ramps are one lane wide. 



G-1 



NOT TO SCALE 




1-81 



Figure G-1. Schematic of the Interstate 80 and 81 Interchange 



G-2 



The 1975 projected ADT figures which were used to design the inter- 
change are shown in Figure G-2. Using the appropriate methods contained 
in the Highway Capacity Manual (1965), the thirtieth highest hourly vol- 
ume was estimated for each traffic movement. These estimates are the 
numbers in parentheses In Figure G-2, 

Critical Analysis of the Existing Design Features 

Deceleration Lane Shape 

Either type of deceleration lane, parallel or tapered, will work 
adequately if designed properly. However, with normal freeway volumes 
and geometry, a tapered deceleration lane is regarded as the optimal 
design because drivers will utilize the lane more effectively since the 
taper conforms to the path that they desire to follow. Therefore, the 
amount of unused pavement area is minimized. Parallel deceleration lanes 
are recommended where volumes are high or where the geometries of the 
exit are less than ideal. The primary advantage of a parallel lane is 
the target value the "stub" provides. 

Table G-1 presents data on the length and shape of each deceleration 
lane for the interchange under study. Both types of deceleration lane 
have been used In the interchange. 

A parallel deceleration lane should be used where sight distance to 
the exit gore is restricted by either horizontal or vertical curvature. 
The sight distances to the exit gores of this interchange are all adequate 
except on the approach to Ramp H. Here, as is shown in Figure G-3, the 
mainline roadway is curving to the right such that the exit ramp gore is 
hidden from the view of the approaching driver by a side slope. Good usage 
has been made of the target value provided by a parallel lane. Despite 

G-3 



9700 
(1494L 



NOT TO SCALE 




7524 7524 
(1159) (1159) 



Figure G-2. Projected Traffic Volumes (VPH) 



P-4 



TABLE G-1 
DECELERATION LANE SHAPE AND LENGTH 













0) 


■T3 14-1 












CO 


x) 












hJ 






ti 






- 


C ^ 


i M 




o 




a /-> 


fx^^ 


O • 


a n 




•H Q) 




60 J3 


e ^ 


•H U 


O <U 


fi* 


4J p. 




•H Q. 


CO fl. 


•U IW 


O h4 


H <H 


to «0 




CO 


Pi a 


cO ^-^ 


<u 


5 2 


}H x: 




0)0 


>—^ 


^ 


Pi s 


« c 


<U M 




P 


TJ 


<U rCj 


3 


•H 


I-l 




-d 


0) T3 


H -U 


o g 




(U 0) 




P. <u 


+J (U 


<U 60 


m -H 


i{ s 




B 0) 


CO (U 


O C 


en c! 




0) CO 




M en 


O P, 

PL, M 


0) (U 


^g 


A 


Tapered 




32 


'25 


413 


550 


B 


Tapered 




50 


40 


486 


425 


C 


Parallel 


auxiliary lane 


30 


25 


1250 


550 


D 


Tapered 




30 


25 


501 


550 


E 


Parallel 




60 


40 


1045 


425 


F 


Parallel 




50 


40 


1045 


425 


G 


Parallel 


auxiliary lane 


25 


25 


785 


550 


H 


Parallel 




45 


40 


1052 


425 



G-5 




Figure G-3o Northbound Approach, to Exit Ramp H 



G-6 



the fact that the exit gore cannot be seen, the abrupt beginning of the 
parallel lane gives the approaching motorist a definite cue of the im- 
pending exit. 

A parallel deceleration lane should also be employed if an exit 
departs tangentially from the mainline. As is shown in Figure G-A, 
Ramp F, although not tangential, departs on the outside of Interstate 
81 which curves to the left in the vicinity of the exit terminal. Here 
a full width parallel deceleration lane has been provided to better 
define the downstream roadway geometry and give the driver an unmistak- 
able cue that an exit ramp is ahead. In this situation, a parallel 
speed-change lane is more effective in guiding drivers along their proper 
course than a tapered design. 

A parallel auxiliary lane has also been rightly provided adjacent 
to Interstate 81 between the entrance terminal of Ramp G and the exit 
terminal of Ramp C, and between the entrance terminal of Ramp A and the 
exit terminal of Ramp G on Interstate 80. The additional lane improves 
weaving operations between the terminals of the two loop ramps without 
the addition of collector-distributor roads by providing extra pavement 
width in which to accomplish the necessary lane changes. 

The recommendations in the main report suggest that a parallel decel- 
eration lane should be used if ramp curvature is such that the off-ramp 
cannot be safely negotiated at a speed of 40 miles per hour or more. With 
a tapered type of design a driver does not have an auxiliary full width 
lane for the entire deceleration distance. Consequently, it is more 
likely that some braking will occur while a portion of the exiting 
vehicle is still in the freeway mainline. 



G-7 




Figure G-4. Southbound Approach, to Ramp F 



G-8 



The speed at which each ramp can be safely negotiated as well as the 
deceleration lane shape and length for each exit terminal are summarized 
in Table G-1. An examination of the Table indicates that Ramps A and D, 
as shown in Figures G-5 and G-6 respectively, are the only terminals which 
employ a tapered deceleration lane for ramp speeds of less than 40 miles 
per hour. In addition, the Table indicates that both of these decelera- 
tion lanes have shorter lengths than the minimums recommended by the 
AASHO "Blue Book" (1965) and "Red Book" (1957), Therefore, it is theo- 
rized that these terminals will probably experience operational problems. 
Indeed, the run-over post delineators shown in the lower portion of 
Figure G-6 indicate that some difficulties have already arisen at the exit 
terminal for Ramp D. Therefore, it is recommended that the deceleration 
lanes of Ramps A and D should be lengthened minimums of 137 and 49 feet 
respectively. Furthermore, along with the length extensions, the shape 
of the speed-change lanes should be altered to provide a parallel-type 
design. 

Left -Hand Exit 

As is seen in Figure G-2, the interchange under study incorporates 
a left-hand exit from the eastbound roadway of Interstate 80 to serve 
traffic which desires to go northbound on Interstate 81. 

In general, the design community is strongly opposed to the use of 
left-hand exits since operational problems are unavoidable even with well- 
designed left-hand facilities. Nevertheless, left-hand exits cannot be 
excluded from the realm of possible design alternatives because in some 
circumstances a left-hand exit may be the least objectionable design. 

At this interchange, the primary consideration which justifies the 
use of a left-hand exit is the relatively large left turn volume. 



G-9 




Figure G-5o Existing Deceleration Lane for Eamp A 



G-10 





Figure G-60 Existing Deceleration Lane for Ramp D 



G-11 



Figure G-2 shows that approximately 36% of the total volume approaching 
the interchange from the west desires to turn left. In the preceding 
chapters, it is reported that approximately 75% of the design engineers 
would "occasionally" use a left-hand diverging configuration if the left 
turn volume is 50% of the total approach volume. If the left turn vol- 
ume is 30% of the approach volume, only one-third of the engineers sur- 
veyed would possibly employ a divergence from the left side of the free- 
way. Most of the engineers indicated they would design the left- turning 
facility as a major fork. 

Considering absolute volume alone, a loop ramp could have been used 
to accommodate the necessary left turn. Apparently the designers of 
•"his interchange believed that the relatively high left turning volume 
should be handled in a directional manner. A semi-directional ramp 
departing from the right-hand side of the freeway could adequately handle 
the projected volume without the inherent difficulties of a left-side 
facility. But with the three other quadrants of the interchange having a 
clover leaf configuration, the use of a semi-directional ramp requires 
that many additional structures be built. Therefore, only a loop ramp or 
a left-hand facility could be economically justified. 

With a left-hand exit, diverging drivers are required to change 

lanes to the left in order to position their vehicles in the left-hand, 

high-speed freeway lane in advance of the exit gore. For safe freeway 

operations the gradient of the freeway in advance of the exit gore should 

be such that diverging trucks can make the required lane changes at a 

high speed. Furthermore, since the exiting trucks will be traveling in 

the left-hand freeway lane, the approach grades should not be such that 

these trucks would be required to travel at speeds lower than those of 

passenger vehicles. 

G-12 



Figure G-7 shows the horizontal and vertical alignment of Inter- 
state 80 in the vicinity of the left-hand exit. The gore of the left- 
hand exit is preceded by 1,594 feet of -1.3% grade. In advance of this, 
a 5,800 foot, -i- 2,0% grade exists. The first directional sign which 
notifies motorists of the left-hand exit is 1.4 miles upstream of the 
left-hand exit- Therefore, the majority of the required lane changes 
will take place on che +2,0% grade. The "Blue Bcok" (1965) specifies 
that the critical lengch of +2.0% grade for a 15 mile per hour speed 
reduction is 3,000 teet. Since the length of +2,0% grade in advance of 
the exit is greater than 3,000 feet, trucks will be traveling at speeds 
significantly lov^?er than passenger vehicles as they approach the left- 
hand exit. Therefore, the possibility of rear-end collisions is great o 
Recent accident records do not indicate that this approach is hazardous, 
but this may be due to the low volumes which are presently using the 
facility. As volumes increase, problems may very well arise. 

Adequate sight distance is essential if a left-hand exit is to 
operate safely. The workshop discussions and a review cf research 
literature indicate that an approaching driver should be able to clearly 
see the left-hand off-ramp geometry from a point at least one-quarter 
mile upstream from the beginning of the deceleration lane. When the 
driver is even wlch the beginning of the deceleration lane, researchers 
at Northwescern University (1969) believe that he should be able to see 
the mainline pavement for a distance of 1,000 feec beyond the gore and 
the ramp pavement for a distance of 500 feet beyond the gore. 

An investigation of Figure G-7 reveals that, by these standards, 
sight distance is clearly inadequate at this left-hand exit. Although 
a driver can see the mainline pavement for 1,000 feet beyond the gore 
and the ramp pavement for 500 feet beyond the gore when at the beginning 



G-13 



r^ LTv 




5 T 



UJ 



lU 



O 






O 

lA 

d 
1 



S < 

o o 



§8 



ILl 



LA 

o 


U- 


—1 
< 


+ 


o 


O 


r^ - 




to 


xS> Q 


§ 






HH 




^ II 


1- 




Q. _l 


O 

UJ 





Q. 



o 

CO 

CO 

M. 

o 

00 
H 

M; 
O 

«4-r 



cf 
S 
g, 



I 

o 

00 



iLl 

i 

I— I 

_J 
< 

_1 

N 

l-H 



o 

o 



i-L 



LLI 
O 

l-H 

_1 
< 



< 

o 

1—1 

LU 
> 



(N 
ID 



G-14 



il 



of the deceleration lane, the requirement that he be able to see the exit 
gore from a distance of one-quarter mile upstream of the beginning of the 
deceleration lane is not satisfied. As a result of the 900 foot crest 
vertical curve in advance of the speed-change lane, an approaching driver 
can see the ramp gore from a distance of only 550 feet in advance of the 
start of the deceleration lane. 

Signing is the most critical factor influencing left-hand exit ramp 
operations. At this interchange signing is even more important since 
the sight distance to the ramp in question does not satisfy the minimum 
standards recommended in the research literature. 

The present signing sequence for the left-hand exit is shown in 
Figure G-8. Figure G-9 shows the signing plan that has been recommended 
for left-hand exits by Northwestern University (1969). The existing and 
the recommended signing plans generally conform up to the one-quarter 
mile point. Here the researchers at Northwestern recommended a sign 
which gives the drivers a further indication of the left-hand exit and 
also informs them cf the speed at which they should exit onto the decel- 
eration lane- The exiting speed indicated on the sign is approximately 
the same as the posted speed of the highway. This sign is provided in 
order to encourage the exiting drivers to do all cf their decelerating 
while in the deceleration lane and not while m the high-speed, left- 
hand freeway lane. After the driver is in the deceleration lane, the 
advisory signs at the ramp gore will inform him of che deceleration which 
is required to sarely negotiate the ramp curve. Therefore, it is recom- 
mended that the signs shown in Figurfe G-10 be provided along the left-hand 
side of Interstate 80 one-quarter mile upstream of the off-ramp nose. 
These signs will give drivers an additional indication cf the left-hand 
exit and also will encourage high-speed exits. 



G-15 






A. 2,4 miles from the left-hand exit gore 



Figure G-80 The Existing Signing for the Left-Hand Exit 



G-16 




B, lA miles from the left-hand exit gore 




C, At Ramp B or 0.4 mile from the left-hand exit gore 



Figure G-80 (Continued) 



G-17 




Do Left-hand exit gore 



Figure G-8„ (Continued) 



G-18 







r 


1- 


(U 


Q. 


I — 1 


LJJ 


2: 


X 


LU 




LJJ 


Q. 







CO >S> 











H 


D. Q 


Cl. 


5 LU 


2: 


< LU 




(2 Q. 





CO 


hO 




II 





S\ 3 LL 







1 — 1 






2 




^ 


CN 


O"^^ 


H 




y — \ 


(O 


1- 


fe/ 


* 
CO 


Ll_ 




lU 


1- 

1 — 1 

X 

LU 



>- 



g}i 



• X 

LU LU 



X 

w 

C 
(d 

4-1 

<u 
i-I 

u 
o 

«H 

C 
CO 
tH 
Pk 

bO 

C 
•H 

C 

00 
•H 

-a 

(U 
4J 
CO 

<u 

bO 
00 

en 



I 
o 

(1) 

3 



G-19 



VBy NORTH 
WILKES BARRE 




EXIT 
SPEED 
65 MPH 







Figure G-10. Recommended Supplementary Signs 



G-20 



Furthermore, it is recommended that yellow "EXIT ONLY" or "THIS 
LANE MUST EXIT" tabs be placed on the overhead sign at the exit gore 
so that drivers will not mistake the relatively long, parallel decelera- 
tion lane which has been provided in advance of the left-hand exit for 
an added through lane. 

The left-hand exit ramp terminal itself is relatively well-designed. 
The vast majority of design engineers believe that a parallel decelera- 
tion lane should be provided at left-hand exits because of the target 
value it provides. The abrupt full lane width will alert drivers of the 
impending exit and will inform them of the parallel lane which has been 
provided for their use in diverging. Such a deceleration lane configura- 
tion has been provided at this left-hand exit. 

The delineation of both the ramp and the mainline roadway is excel- 
lent. Currently, the terminal area is not lighted » In the future, light- 
ing should be considered if operations in the vicinity of the exit ter- 
minal become hazardous. 

The recovery taper at the left-hand exit ramp nose is 125 feet long. 
All other exit terminals in the interchange have 150 foot tapers. It 
appears somewhat illogical to provide a shorter recovery area at the gore 
of the left-hand exit since this is probably the one location where the 
most driver confusion will result. Using the general procedure outlined 
in the AASHO "Blue Book" (1965), it was found that a 150 foot taper is 
recommended for any exit ramp departing from a 70 mile per hour approach 
roadway with a 10 foot nose offset. Northwestern University (1969) has 
suggested that a minimum taper length of 250 feet should be used for left- 
hand exits. Therefore, it is evident that the present recovery area is 
inadequate, and thus it should be extended to provide a minimtim taper 

length of 250 feet, 

G-21 



Single Versus Double Exit Design 

Research literature and the results of the workshop sessions indi- 
cate that the majority of design engineers believe that single exit 
interchange should be used as often as possible for freeway-to-freeway 
interchanges. Since the majority of interchanges along a route are 
likely to be single exit, diamond-type designs, a single exit interchange 
generally assures uniform exiting patterns. With a single exit pattern 
signing is simplified since points of decision are separated. Further- 
more, with cloverleaf-type designs, the provision of single exits with 
the concomitant collector-distributor roadways also removes weaving areas 
from the mainline roadway. 

Despite the operational advantages which can be gained by a single 
exit design, experienced designers recognize that such configurations 
cannot always be economically justified due to the additional costs 
which would be required in order to furnish the collector-distributor 
roads and the wider and/or additional structures they would require. 
Designs with two exits on an approach can operate adequately. Therefore, 
the provision of a single exit can be viewed as a desirable, but not 
essential, operational refinement. 

The interchange under study is a conventional cloverleaf with the 
exception of a left-hand exit on one approach. As such, each approach 
roadway has two exits from the mainline. Almost all of the interchanges 
located along each approach to the interchange under study possess single 
exits from the mainline. This suggests that the Interstate 80 and 81 
interchange should likewise be of the single exit type so that the 
design would conform to the pattern that drivers would expect to encounter. 

Nevertheless, the interchange was designed as a two-exit conventional 
cloverleaf with a left-hand exit on one approach. With the left-hand exit, 



G-22 



single exits could be provided only on the southbound and westbound 
approaches. Considering the very low projected traffic volumes and the 
rural location, a basic cloverleaf configuration is the least expensive 
configuration which could be employed. As a result of the low volumes, 
each of the exit terminals of the present two-exit design is operating 
at level of service A. Thus, the author believes that the designers 
were justified in employing successive exits on each approach. 

If successive exits are used, a driver must be given adequate dis- 
tance to make decisions and maneuver between the two exit ramp gores. 
Table G-2 compares the distances between the exit gores in the inter- 
change with those which have been recommended as a result of this study. 
The recommended distances in Table G-2 were chosen because at least 
50% of the design experts at the workshops endorsed a distance which 
fell within the specified range. The Table indicates that the distances 
between the two exits on each approach to the interchange far exceed 
those recommended. 

Suggested Design Improvements 

A review of relevant research literature and the results of a 
study of the opinions of a representative sample of design experts indi- 
cate that the following improvements should be made to improve the 
safety and operations at the exit terminals of this major interchange: 

1. The deceleration lanes for Ramps A and D should be lengthened 
minlmums of 137 and 49 feet respectively and should be changed 
to a parallel-type design. 

2. A supplementary sign, with the message shown in Figure G-10, 
should be provided along the left-hand side of the roadway one- 
quarcer mile in advance of the left-hand exit gore. 



G-23 



TABLE G-2 
A COMPARISON OF THE EXISTING AND RECOMMENDED 
DISTANCES BETWEEN SUCCESSIVE EXITS 



Direction 



Exit 
Terminal 



Eastbound 


B-E 


Westbound 


D-G 


Northbound 


H-A 


Southbound 


F-C 



Existing 
Distance (ft.) 



2,095 
2,708 
2,242 
2,260 



Recommended Distance 
(ft.) 



Minimum 
500-800 



Desirable 
1,000-1,200 



800-1,000 1,000-1.500 
800-1,000 1,000-1,500 
800-1,000 1,000-1,500 



G-24 



3. Yellow "EXIT ONLY" or "THIS LANE MUST EXIT" tabs should be 
added to the overhead sign at the left-hand exit gore. 

4. The tapered recovery area at the left-hand exit gore should 
be lengthened to a minimum of 250 feet. 






G-25 



H* 



DOT LIBRARY 



DDmt.l7S 




R&D